---
schema_version: 7
rubric_version: "0.3"
canonical_url: "https://podium.do/compare/agents/products/mini-swe-agent/"
data_url: "https://podium.do/compare/agents/products/mini-swe-agent/data.json"
generated_at: 2026-11-22
oldest_checked_at: 2026-11-16
last_checked_at: 2026-11-16
publisher: "Podium"
---

# mini-SWE-agent: ADE findings

> Podium publishes this catalog and sells an agent development environment. Podium chooses the products, criteria, sources and wording, so treat these pages as vendor-authored research. Check important claims against the linked primary sources. Sourced findings follow the same evidence rules for Podium and its competitors. Facts labeled Podium team research come from Podium's own team and have no linked source. None of this removes the conflict of interest.

> Based on data from adecomparison.com by Podium (https://adecomparison.com; source repository https://github.com/madeinorbit/ade-comparison, snapshot 4322f0c of 2026-11-22), used under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Source links and check dates are kept. Podium arranges the findings into pairs, applies its publication checks, and added its own research where marked "Podium team research". Data license terms: https://github.com/madeinorbit/ade-comparison/blob/4322f0c035f303791413639d1462a138443d8c1e/DATA-LICENSE.md

51 of 51 criteria have a sourced finding; 51 were also cross-checked by a second agent. Source checks: 2026-11-16. A missing sourced finding is not evidence that a capability is absent.

- Website: https://mini-swe-agent.com
- Canonical HTML: https://podium.do/compare/agents/products/mini-swe-agent/
- Machine-readable JSON: https://podium.do/compare/agents/products/mini-swe-agent/data.json
- Rubric: https://podium.do/compare/agents/rubric/index.md
- Comparisons with mini-SWE-agent on podium.do: 1 (listed in the JSON and HTML)
- Compare mini-SWE-agent with any other product on adecomparison.com by Podium: https://adecomparison.com/compare/?products=mini-swe-agent

## Access and agent choice

### ACC-01: Client access

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Python package installable with pip on any OS: the mini REPL CLI, mini-extra batch and config commands, and a Textual trajectory inspector. No desktop app, mobile app, browser UI or editor extension. Bubblewrap isolation is Linux only.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- desktop_os: Packaged as an OS-independent Python wheel (no native desktop client); bubblewrap environment is Linux only.
- mobile_apps: No mobile app; bounded search of repo and docs.
- browser: No browser UI in src or docs.
- mobile_browser: No browser UI.

- desktop_os — Desktop os [partial]: Packaged as an OS-independent Python wheel (no native desktop client); bubblewrap environment is Linux only. Sources: mini-swe-agent:pyproject, mini-swe-agent:envdoc.
- mobile_apps — Mobile apps [unsupported]: No mobile app; bounded search of repo and docs. Sources: mini-swe-agent:search.
- browser — Browser [unsupported]: No browser UI in src or docs. Sources: mini-swe-agent:search.
- mobile_browser — Mobile browser [unsupported]: No browser UI. Sources: mini-swe-agent:search.
- cli — Cli [supported]: mini, mini-swe-agent, mini-extra commands. Sources: mini-swe-agent:pyproject, mini-swe-agent:mini.
- editor_extensions — Editor extensions [unsupported]: No editor extension in repo (searched vscode, extension, plugin). Sources: mini-swe-agent:search.
- task_actions — Task actions [partial]: Start, confirm or reject each command, switch modes and interrupt in the terminal; review by reading output; resume is not supported except by rerunning with the saved trajectory path. Sources: mini-swe-agent:interactive, mini-swe-agent:minidoc.
- notifications — Notifications [unsupported]: No notification code found. Sources: mini-swe-agent:interactive.
- cross_device_continuity — Cross device continuity [unsupported]: Runs in a terminal; trajectories are local files. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:pyproject, mini-swe-agent:envdoc, mini-swe-agent:search, mini-swe-agent:mini, mini-swe-agent:interactive, mini-swe-agent:minidoc, mini-swe-agent:output

### ACC-02: Agent and model choice

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Single embedded minimal agent whose only tool is bash; models via litellm (most providers), OpenRouter, Portkey, Requesty, text-based or tool-call formats, local models; custom agent, model and environment classes by dotted path. It does not drive other coding-agent CLIs.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), Python library (agent, model and environment classes); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- harnesses: One embedded agent; no third-party agent CLI is launched.
- mixed_harness_operation: Only one harness.

- harnesses — Harnesses [partial]: One embedded agent; no third-party agent CLI is launched. Sources: mini-swe-agent:readme, mini-swe-agent:agent.
- model_providers — Model providers [supported]: litellm, OpenRouter, Portkey, Requesty model classes. Sources: mini-swe-agent:modelsdoc, mini-swe-agent:readme.
- custom_adapters — Custom adapters [supported]: --model-class, --agent-class, --environment-class accept dotted import paths. Sources: mini-swe-agent:mini.
- session_integration — Session integration [supported]: Embedded loop with message history in memory. Sources: mini-swe-agent:agent.
- mixed_harness_operation — Mixed harness operation [unsupported]: Only one harness. Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:agent, mini-swe-agent:modelsdoc, mini-swe-agent:mini, mini-swe-agent:search

### ACC-03: Subscriptions, API keys and vendor credits

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

API keys through litellm environment variables or the .env config, local models, OpenRouter and Portkey gateways; no vendor credits and no subscription-login flow in the code.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- agent_subscriptions: No OAuth or subscription login code (searched oauth, login, subscription in src).
- account_switching: Rotating keys for parallel runs described in global config docs.
- provider_restrictions: Provider terms are third-party.

- agent_subscriptions — Agent subscriptions [unsupported]: No OAuth or subscription login code (searched oauth, login, subscription in src). Sources: mini-swe-agent:modelsdoc.
- api_accounts — Api accounts [supported]: API keys per provider. Sources: mini-swe-agent:modelsdoc.
- vendor_credits — Vendor credits [not-applicable]: No vendor credits. Sources: mini-swe-agent:readme.
- local_models — Local models [supported]: Local model guide and litellm local providers. Sources: mini-swe-agent:modelsdoc.
- account_switching — Account switching [partial]: Rotating keys for parallel runs described in global config docs. Sources: mini-swe-agent:globalconfig.
- vendor_account_requirement — Vendor account requirement [supported]: No account with the project required. Sources: mini-swe-agent:readme.
- provider_restrictions — Provider restrictions [unknown]: Provider terms are third-party.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:modelsdoc, mini-swe-agent:readme, mini-swe-agent:globalconfig

### ACC-04: Onboarding and project prerequisites

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

pip install, mini-extra config setup for a model and key, then run mini in any directory; no Git requirement; environments (docker, singularity, bubblewrap) need their own tooling. No managed cloud or VPS onboarding.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- hosted_git: Opening PRs is left to the model through bash and gh.
- managed_cloud_setup: No managed cloud.
- byo_vps_onboarding: No VPS onboarding flow.

- first_task_setup — First task setup [supported]: Install, mini-extra config setup, run mini. Sources: mini-swe-agent:modelsdoc, mini-swe-agent:minidoc.
- no_git — No git [supported]: Runs in the current directory. Sources: mini-swe-agent:minidoc.
- local_git_without_remote — Local git without remote [supported]: Git is just shell commands for the model. Sources: mini-swe-agent:readme.
- hosted_git — Hosted git [partial]: Opening PRs is left to the model through bash and gh. Sources: mini-swe-agent:readme.
- prerequisite_guidance — Prerequisite guidance [supported]: Setup guides and environment prerequisites documented. Sources: mini-swe-agent:envdoc, mini-swe-agent:modelsdoc.
- managed_cloud_setup — Managed cloud setup [unsupported]: No managed cloud. Sources: mini-swe-agent:search.
- byo_vps_onboarding — Byo vps onboarding [unsupported]: No VPS onboarding flow. Sources: mini-swe-agent:envdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:modelsdoc, mini-swe-agent:minidoc, mini-swe-agent:readme, mini-swe-agent:envdoc, mini-swe-agent:search

### ACC-05: Programmable access

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

CLI (mini, mini-extra), a Python API documented in python_bindings and the reference, YAML configs; no HTTP API, MCP server or webhooks.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), Python library (agent, model and environment classes), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- api: No HTTP server or API in src.
- mcp_server: No MCP code in src or docs (searched mcp).
- webhooks: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- operations: Run, set limits and read trajectories; no remote steering or stop API.

- cli — Cli [supported]: mini and mini-extra. Sources: mini-swe-agent:pyproject, mini-swe-agent:extra.
- api — Api [unsupported]: No HTTP server or API in src. Sources: mini-swe-agent:search.
- sdk — Sdk [supported]: Python classes for agent, model and environment. Sources: mini-swe-agent:readme, mini-swe-agent:agent.
- mcp_server — Mcp server [unsupported]: No MCP code in src or docs (searched mcp). Sources: mini-swe-agent:search.
- webhooks — Webhooks [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- operations — Operations [partial]: Run, set limits and read trajectories; no remote steering or stop API. Sources: mini-swe-agent:agent, mini-swe-agent:output.
- authentication — Authentication [not-applicable]: No server surface. Sources: mini-swe-agent:readme.
- stability — Stability [partial]: v2 changed output and APIs with a migration guide; pre-3.0 library. Sources: mini-swe-agent:rel.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:pyproject, mini-swe-agent:extra, mini-swe-agent:search, mini-swe-agent:readme, mini-swe-agent:agent, mini-swe-agent:output, mini-swe-agent:rel

### ACC-06: External tools and extensions

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No MCP client, plugins, skills or instruction-file import: the agent uses bash plus Jinja-templated system and instance prompts configured in YAML; behavior is extended by subclassing in Python.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), Python library (agent, model and environment classes); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- mcp_client: No MCP client code or docs.
- plugins: No plugin system; subclass or dotted-path classes.
- custom_tools: Bash is the only tool by design.
- instructions_import: No AGENTS.md or CLAUDE.md import found; prompts come from YAML config.

- mcp_client — Mcp client [unsupported]: No MCP client code or docs. Sources: mini-swe-agent:search.
- plugins — Plugins [unsupported]: No plugin system; subclass or dotted-path classes. Sources: mini-swe-agent:mini.
- custom_tools — Custom tools [unsupported]: Bash is the only tool by design. Sources: mini-swe-agent:search.
- instructions_import — Instructions import [unsupported]: No AGENTS.md or CLAUDE.md import found; prompts come from YAML config. Sources: mini-swe-agent:agent.
- authentication — Authentication [not-applicable]: No integrations. Sources: mini-swe-agent:readme.
- permission_scope — Permission scope [partial]: Confirm mode with regex whitelist for commands. Sources: mini-swe-agent:interactive.
- integration_owner — Integration owner [not-applicable]: No integrations. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:mini, mini-swe-agent:agent, mini-swe-agent:readme, mini-swe-agent:interactive

### ACC-07: Client implementation

Verification status: Agent cross-checked (agent-cross-checked)
Support: supported

Python terminal application (prompt_toolkit and rich REPL, Textual inspector); no desktop or mobile clients.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- desktop_implementation — Desktop implementation [supported]: Python terminal UI with rich, prompt_toolkit and Textual. Sources: mini-swe-agent:pyproject, mini-swe-agent:inspector.
- mobile_implementation — Mobile implementation [not-applicable]: No mobile client. Sources: mini-swe-agent:readme.
- hybrids — Hybrids [not-applicable]: None. Sources: mini-swe-agent:readme.
- implementation_evidence — Implementation evidence [supported]: pyproject.toml and sources at the pinned tag. Sources: mini-swe-agent:pyproject.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:pyproject, mini-swe-agent:inspector, mini-swe-agent:readme

## Planning and context

### PLAN-01: Intake and external tracker relationship

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Intake is a typed or flag-provided task string; batch mode ingests benchmark instances; no tracker integration.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- external_import: Benchmark datasets in batch mode only.
- chat_git_triggers: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- stable_ids: Benchmark instance IDs only.
- sync: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- local_creation — Local creation [supported]: Task typed or passed with -t. Sources: mini-swe-agent:mini.
- prompt_intake — Prompt intake [supported]: Prompt or file. Sources: mini-swe-agent:mini.
- external_import — External import [partial]: Benchmark datasets in batch mode only. Sources: mini-swe-agent:swebench.
- chat_git_triggers — Chat git triggers [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- authoritative_tracker — Authoritative tracker [not-applicable]: No tracker. Sources: mini-swe-agent:readme.
- stable_ids — Stable ids [partial]: Benchmark instance IDs only. Sources: mini-swe-agent:swebench.
- sync — Sync [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- supported_edits — Supported edits [not-applicable]: None. Sources: mini-swe-agent:readme.
- external_roundtrips — External roundtrips [not-applicable]: None. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:mini, mini-swe-agent:swebench, mini-swe-agent:search, mini-swe-agent:readme

### PLAN-02: Durable task plans

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No plan feature: the model reasons in its THOUGHT section before each command; the whole conversation is the only persistent record.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- plans: No plan objects.
- acceptance_criteria: None.
- editors: None.
- revisions: None.

- plans — Plans [unsupported]: No plan objects. Sources: mini-swe-agent:agent.
- acceptance_criteria — Acceptance criteria [unsupported]: None. Sources: mini-swe-agent:agent.
- editors — Editors [unsupported]: None. Sources: mini-swe-agent:agent.
- revisions — Revisions [unsupported]: None. Sources: mini-swe-agent:agent.
- implementation_links — Implementation links [unsupported]: None. Sources: mini-swe-agent:agent.
- amendment — Amendment [partial]: User can interrupt and add messages in the REPL. Sources: mini-swe-agent:interactive.
- restart_persistence — Restart persistence [partial]: Trajectory saved to .traj.json; resuming a run from it is not documented. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:agent, mini-swe-agent:interactive, mini-swe-agent:output

### PLAN-03: Hierarchy, dependencies and evolving scope

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No hierarchy, dependencies or epics; a single linear message history.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- epics: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- hierarchy: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- blockers: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- readiness: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- epics — Epics [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- hierarchy — Hierarchy [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- blockers — Blockers [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- readiness — Readiness [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- enforcement — Enforcement [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- state_propagation — State propagation [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- task_to_epic — Task to epic [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- identity_retention — Identity retention [not-applicable]: None. Sources: mini-swe-agent:readme.
- workflow_constraints — Workflow constraints [supported]: No required workflow. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:readme

### PLAN-04: Shared context and memory

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Behavior is set by YAML system and instance templates; there is no memory, skills or repository-instruction import; each run starts from the task prompt.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), Python library (agent, model and environment classes); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- instructions: YAML templates for system and instance prompts.
- skills: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- project_docs: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- memory: None found (searched memory in src).

- instructions — Instructions [partial]: YAML templates for system and instance prompts. Sources: mini-swe-agent:agent.
- skills — Skills [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- project_docs — Project docs [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- memory — Memory [unsupported]: None found (searched memory in src). Sources: mini-swe-agent:search.
- scope — Scope [partial]: Global versus per-config files. Sources: mini-swe-agent:globalconfig.
- provenance — Provenance [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- correction — Correction [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- deletion — Deletion [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- freshness — Freshness [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- takeover_continuity — Takeover continuity [unsupported]: Only the local trajectory file survives. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:agent, mini-swe-agent:search, mini-swe-agent:globalconfig, mini-swe-agent:output

### PLAN-05: Architecture and API awareness

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No search, indexing or knowledge tools; the agent uses bash commands the model chooses (grep, find, etc.).

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- code_search: Only via shell commands the model runs.
- indexing: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- external_knowledge: Only via curl or similar in bash.
- multimodal_inputs: openai_multimodal utilities exist for model messages; no UI image input found.

- code_search — Code search [partial]: Only via shell commands the model runs. Sources: mini-swe-agent:readme.
- indexing — Indexing [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- external_knowledge — External knowledge [partial]: Only via curl or similar in bash. Sources: mini-swe-agent:readme.
- multimodal_inputs — Multimodal inputs [partial]: openai_multimodal utilities exist for model messages; no UI image input found. Sources: mini-swe-agent:model.
- freshness — Freshness [not-applicable]: No index. Sources: mini-swe-agent:readme.
- access_scope — Access scope [partial]: Whatever the environment exposes. Sources: mini-swe-agent:envdoc.
- api_retrieval — Api retrieval [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- citations — Citations [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- conflict_detection — Conflict detection [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- review_enforcement — Review enforcement [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:search, mini-swe-agent:model, mini-swe-agent:envdoc

### PLAN-06: Native issue management and idea backlog

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No issue tracker.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- create: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- edit: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- discuss: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- prioritize: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- create — Create [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- edit — Edit [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- discuss — Discuss [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- prioritize — Prioritize [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- label — Label [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- assign — Assign [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- search — Search [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- filter — Filter [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- archive — Archive [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- reopen — Reopen [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- project_epic_views — Project epic views [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- idea_parking — Idea parking [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- explicit_dispatch — Explicit dispatch [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- task_session_separation — Task session separation [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- metadata — Metadata [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- durable_states — Durable states [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- external_tracker_requirement — External tracker requirement [supported]: An external tracker is needed for any issue management. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:readme

### PLAN-07: Agent-native issue operations

Verification status: Agent cross-checked (agent-cross-checked)
Support: unsupported

No shared issue records for agents.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- shared_ids — Shared ids [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- read — Read [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- create — Create [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- update — Update [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- claim — Claim [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- assign — Assign [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- state_changes — State changes [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- dependencies — Dependencies [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- children — Children [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- results — Results [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- human_visibility — Human visibility [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- attribution — Attribution [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- permissions — Permissions [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- conflicts — Conflicts [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search

### PLAN-08: Living product specifications

Verification status: Agent cross-checked (agent-cross-checked)
Support: unsupported

No living specifications.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- persistent_requirements — Persistent requirements [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- human_edits — Human edits [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- agent_edits — Agent edits [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- revisions — Revisions [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- approvals — Approvals [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- implementation_links — Implementation links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- later_retrieval — Later retrieval [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- drift_detection — Drift detection [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- review_integration — Review integration [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search

## Coordination and automation

### COORD-01: Parallel work

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

A single agent per mini run; batch benchmark mode runs many instances in parallel worker threads with a live progress table showing step, cost and elapsed time per instance.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench), mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- concurrent_sessions: Multiple processes can be started manually; batch threads in swebench.
- quotas: Worker count is the only limit.
- observed_concurrency: Not measured here.

- concurrent_sessions — Concurrent sessions [partial]: Multiple processes can be started manually; batch threads in swebench. Sources: mini-swe-agent:swebench.
- concurrent_tasks — Concurrent tasks [supported]: --workers parallel instances. Sources: mini-swe-agent:swebench.
- quotas — Quotas [partial]: Worker count is the only limit. Sources: mini-swe-agent:swebench.
- observed_concurrency — Observed concurrency [unknown]: Not measured here.
- status_visibility — Status visibility [supported]: Batch progress table shows per-instance step, cost and time. Sources: mini-swe-agent:swebench.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench

### COORD-02: Ownership and dispatch

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No assignment, claims or routing; batch workers pull instances from a list.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- human_assignment: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- agent_assignment: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- claims: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- queues: Batch work list across workers.

- human_assignment — Human assignment [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- agent_assignment — Agent assignment [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- claims — Claims [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- queues — Queues [partial]: Batch work list across workers. Sources: mini-swe-agent:swebench.
- routing — Routing [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- duplicate_prevention — Duplicate prevention [partial]: Batch skips finished instances via output files. Sources: mini-swe-agent:swebench.
- ownership_enforcement — Ownership enforcement [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:swebench

### COORD-03: Inspectable delegation and communication

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No delegation: a single agent with a linear history; no subagents, messaging or cross-harness teams.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- roles: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- model_harness_effort: None; a roulette model switches models between calls.
- messages: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- results_return: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- roles — Roles [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- model_harness_effort — Model harness effort [unsupported]: None; a roulette model switches models between calls. Sources: mini-swe-agent:search.
- messages — Messages [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- results_return — Results return [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- cross_harness — Cross harness [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- orchestration_owner — Orchestration owner [not-applicable]: None. Sources: mini-swe-agent:readme.
- activation — Activation [not-applicable]: None. Sources: mini-swe-agent:readme.
- authentication — Authentication [not-applicable]: None. Sources: mini-swe-agent:readme.
- live_progress — Live progress [not-applicable]: None. Sources: mini-swe-agent:readme.
- retained_results — Retained results [not-applicable]: None. Sources: mini-swe-agent:readme.
- followup — Followup [not-applicable]: None. Sources: mini-swe-agent:readme.
- invocation — Invocation [not-applicable]: None. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:readme

### COORD-04: Unattended triggers

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No schedules, triggers or recurring tasks; runs start when a human or script invokes them.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- schedules: No scheduler in src.
- events: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- recurring_tasks: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- completion_conditions: Step, cost and wall-time limits end a run.

- schedules — Schedules [unsupported]: No scheduler in src. Sources: mini-swe-agent:search.
- events — Events [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- recurring_tasks — Recurring tasks [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- completion_conditions — Completion conditions [partial]: Step, cost and wall-time limits end a run. Sources: mini-swe-agent:agent.
- scheduler_location — Scheduler location [not-applicable]: None. Sources: mini-swe-agent:readme.
- overlaps — Overlaps [not-applicable]: None. Sources: mini-swe-agent:readme.
- retries — Retries [not-applicable]: None. Sources: mini-swe-agent:readme.
- client_independence — Client independence [partial]: Batch runs are ordinary processes that can be run headless. Sources: mini-swe-agent:swebench.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:agent, mini-swe-agent:readme, mini-swe-agent:swebench

### COORD-05: Intervention and attention

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Ctrl+C interrupt, confirm or reject each command with a message, mode switching, human takeover, exit confirmation; no pause or remote steering; history saved to the trajectory.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- pause: Interrupt returns to a prompt where the user can resume with a mode choice.
- resume: Continue the same process after interruption; no resume command from a trajectory.
- notifications: None.
- escalation: Format-error and limit exceptions end the run.

- pause — Pause [partial]: Interrupt returns to a prompt where the user can resume with a mode choice. Sources: mini-swe-agent:interactive.
- stop — Stop [supported]: Ctrl+C interrupts; trajectory is saved. Sources: mini-swe-agent:interactive, mini-swe-agent:mini.
- resume — Resume [partial]: Continue the same process after interruption; no resume command from a trajectory. Sources: mini-swe-agent:interactive.
- steer — Steer [supported]: User can add messages and take over. Sources: mini-swe-agent:interactive.
- notifications — Notifications [unsupported]: None. Sources: mini-swe-agent:interactive.
- pending_decisions — Pending decisions [supported]: Confirmation prompts. Sources: mini-swe-agent:interactive.
- escalation — Escalation [partial]: Format-error and limit exceptions end the run. Sources: mini-swe-agent:agent.
- visible_errors — Visible errors [supported]: Errors printed and recorded in the trajectory. Sources: mini-swe-agent:agent, mini-swe-agent:output.
- chat_terminal_parity — Chat terminal parity [not-applicable]: Single terminal surface. Sources: mini-swe-agent:minidoc.
- retained_history — Retained history [supported]: .traj.json. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:interactive, mini-swe-agent:mini, mini-swe-agent:agent, mini-swe-agent:output, mini-swe-agent:minidoc

### COORD-06: Fleet and epic management UI

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Batch progress table and the trajectory inspector; no fleet or epic view.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- fleet_fixture: Batch progress view lists instances with step, cost, time.
- parent_child: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- dependencies: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- attention: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- fleet_fixture — Fleet fixture [partial]: Batch progress view lists instances with step, cost, time. Sources: mini-swe-agent:swebench.
- parent_child — Parent child [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- dependencies — Dependencies [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- attention — Attention [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- history — History [supported]: Trajectory browser. Sources: mini-swe-agent:inspector.
- workload — Workload [partial]: Overall progress, cost and ETA in batch view. Sources: mini-swe-agent:swebench.
- navigation_errors — Navigation errors [partial]: Exit statuses summarized. Sources: mini-swe-agent:swebench.
- fixture_mode — Fixture mode [not-applicable]: None. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench, mini-swe-agent:search, mini-swe-agent:inspector, mini-swe-agent:readme

### COORD-07: Workflow flexibility and orchestration visibility

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No presets or natural-language configuration; behavior is YAML config and CLI flags, with modes confirm/yolo/human switchable in a run.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), Python library (agent, model and environment classes); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- growing_epic: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- presets: Config files such as mini.yaml and benchmark configs.
- natural_language: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- agent_configuration: YAML templates and limits.

- quick_task — Quick task [supported]: mini -t. Sources: mini-swe-agent:mini.
- growing_epic — Growing epic [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- presets — Presets [partial]: Config files such as mini.yaml and benchmark configs. Sources: mini-swe-agent:mini.
- natural_language — Natural language [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- agent_configuration — Agent configuration [partial]: YAML templates and limits. Sources: mini-swe-agent:agent.
- in_run_changes — In run changes [supported]: Switch mode with /c /y /u. Sources: mini-swe-agent:interactive.
- live_roles — Live roles [not-applicable]: No roles. Sources: mini-swe-agent:readme.
- retained_relationships — Retained relationships [not-applicable]: None. Sources: mini-swe-agent:readme.
- discoverability — Discoverability [supported]: Docs and --help. Sources: mini-swe-agent:minidoc.
- auth_friction — Auth friction [supported]: Keys only. Sources: mini-swe-agent:modelsdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:mini, mini-swe-agent:search, mini-swe-agent:agent, mini-swe-agent:interactive, mini-swe-agent:readme, mini-swe-agent:minidoc, mini-swe-agent:modelsdoc

## Execution and reliability

### RUN-01: Laptop-independent execution

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Runs locally by default; docker or podman, Singularity, bubblewrap, ConTree, and SWE-ReX Docker or Modal cloud environments run commands elsewhere while the agent loop stays in your process; no managed vendor cloud and nothing continues after your process ends.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- byo_remote: SWE-ReX and docker hosts configurable; setup is manual.
- managed_cloud: No project-operated cloud; Modal is a third-party service the user brings.
- hybrid: Agent runs locally while commands run in Modal or containers.
- provisioning: Container started per run.

- local — Local [supported]: LocalEnvironment default for mini. Sources: mini-swe-agent:envdoc, mini-swe-agent:local.
- byo_remote — Byo remote [partial]: SWE-ReX and docker hosts configurable; setup is manual. Sources: mini-swe-agent:envdoc.
- self_hosted — Self hosted [not-applicable]: No service. Sources: mini-swe-agent:readme.
- managed_cloud — Managed cloud [unsupported]: No project-operated cloud; Modal is a third-party service the user brings. Sources: mini-swe-agent:envdoc.
- hybrid — Hybrid [partial]: Agent runs locally while commands run in Modal or containers. Sources: mini-swe-agent:envdoc.
- provisioning — Provisioning [partial]: Container started per run. Sources: mini-swe-agent:docker.
- payer — Payer [supported]: User pays infrastructure and models. Sources: mini-swe-agent:readme.
- resources — Resources [partial]: Container options through run_args. Sources: mini-swe-agent:docker.
- region — Region [not-applicable]: Provider dependent. Sources: mini-swe-agent:envdoc.
- expiry — Expiry [supported]: container_timeout default 2h. Sources: mini-swe-agent:docker.
- client_disconnect — Client disconnect [unsupported]: The agent loop is in the client process; it stops when the process ends. Sources: mini-swe-agent:agent.
- other_device_continuation — Other device continuation [unsupported]: None. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:envdoc, mini-swe-agent:local, mini-swe-agent:readme, mini-swe-agent:docker, mini-swe-agent:agent, mini-swe-agent:output

### RUN-02: Workspace and process isolation

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Local environment has no isolation; docker/podman containers, Singularity and bubblewrap (Linux) isolate execution per run; each action is an independent subprocess; no worktrees or branches created.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- Default mini environment is not sandboxed.

- checkout — Checkout [supported]: Local runs in the current directory. Sources: mini-swe-agent:local.
- branch — Branch [unsupported]: No branch handling. Sources: mini-swe-agent:search.
- worktree — Worktree [unsupported]: Bounded search of src found no worktree code. Sources: mini-swe-agent:search.
- container — Container [supported]: DockerEnvironment (docker or podman executable). Sources: mini-swe-agent:docker.
- vm — Vm [partial]: Modal and ConTree sandboxes via SWE-ReX; not VM-specific. Sources: mini-swe-agent:envdoc.
- os_sandbox — Os sandbox [supported]: Bubblewrap on Linux. Sources: mini-swe-agent:bwrap.
- filesystem — Filesystem [partial]: Container or bubblewrap filesystem boundaries. Sources: mini-swe-agent:docker, mini-swe-agent:bwrap.
- process — Process [supported]: Per-action subprocess in the environment. Sources: mini-swe-agent:local, mini-swe-agent:docker.
- network — Network [partial]: Container networking is up to run_args. Sources: mini-swe-agent:docker.
- secrets — Secrets [partial]: forward_env selects forwarded environment variables. Sources: mini-swe-agent:docker.
- code_separation — Code separation [partial]: Per-run container. Sources: mini-swe-agent:docker.
- security_boundary — Security boundary [partial]: Documented as sandboxing options, not hostile-code guarantees. Sources: mini-swe-agent:envdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:local, mini-swe-agent:search, mini-swe-agent:docker, mini-swe-agent:envdoc, mini-swe-agent:bwrap

### RUN-03: Recovery, expiry and continuity

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

The trajectory is saved when a run ends or is interrupted; there is no process recovery; a crash loses the in-memory run apart from the trajectory written so far.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- client_disconnect: Process ends with the terminal.
- worker_crash: Trajectory persisted on exit and exceptions.
- process_reconnect: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- task_resume: No resume command.

- client_disconnect — Client disconnect [unsupported]: Process ends with the terminal. Sources: mini-swe-agent:agent.
- worker_crash — Worker crash [partial]: Trajectory persisted on exit and exceptions. Sources: mini-swe-agent:agent.
- service_restart — Service restart [not-applicable]: No service. Sources: mini-swe-agent:readme.
- expiry — Expiry [supported]: Container timeout. Sources: mini-swe-agent:docker.
- transcript_reload — Transcript reload [supported]: Inspector loads .traj.json. Sources: mini-swe-agent:inspector.
- process_reconnect — Process reconnect [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- task_resume — Task resume [unsupported]: No resume command. Sources: mini-swe-agent:search.
- state_reconstruction — State reconstruction [partial]: Messages can be reloaded manually. Sources: mini-swe-agent:output.
- context_loss — Context loss [partial]: Linear history; long histories are not compacted. Sources: mini-swe-agent:readme.
- duplicate_actions — Duplicate actions [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- manual_repair — Manual repair [partial]: Manual rerun. Sources: mini-swe-agent:output.
- multi_day_continuity — Multi day continuity [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:agent, mini-swe-agent:readme, mini-swe-agent:docker, mini-swe-agent:inspector, mini-swe-agent:search, mini-swe-agent:output

### RUN-04: Development environment reliability

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Environments start containers and forward selected environment variables; no dependency caching, port allocation or setup scripts are provided.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench), mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- dependencies: Benchmark containers come prepared; no setup scripts.
- caches: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- snapshots: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- services: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- dependencies — Dependencies [partial]: Benchmark containers come prepared; no setup scripts. Sources: mini-swe-agent:swebench.
- caches — Caches [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- snapshots — Snapshots [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- services — Services [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- ports — Ports [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- secrets — Secrets [partial]: forward_env. Sources: mini-swe-agent:docker.
- duplicate_installs — Duplicate installs [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- collisions — Collisions [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- remote_setup — Remote setup [partial]: Modal and SWE-ReX configuration. Sources: mini-swe-agent:envdoc.
- automatic_repair — Automatic repair [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench, mini-swe-agent:search, mini-swe-agent:docker, mini-swe-agent:envdoc

### RUN-05: Limits and failure handling

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Step limit, cost limit (default $3), wall-time limit, per-command timeout (30s), container timeout (2h), consecutive format-error limit, model retries with exponential backoff; limits are enforced stops.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- token_limits: No token cap; cost limit instead.
- provider_failures: Retries then error.

- time_limits — Time limits [supported]: wall_time_limit_seconds and container_timeout. Sources: mini-swe-agent:agent, mini-swe-agent:docker.
- turn_limits — Turn limits [supported]: step_limit. Sources: mini-swe-agent:agent.
- token_limits — Token limits [unsupported]: No token cap; cost limit instead. Sources: mini-swe-agent:agent.
- spend_limits — Spend limits [supported]: cost_limit default 3.0. Sources: mini-swe-agent:agent.
- timeouts — Timeouts [supported]: Per-command timeout default 30s. Sources: mini-swe-agent:local.
- retries — Retries [supported]: Tenacity retries with exponential wait. Sources: mini-swe-agent:retry.
- provider_failures — Provider failures [partial]: Retries then error. Sources: mini-swe-agent:retry.
- enforcement — Enforcement [supported]: LimitsExceeded exception ends the run. Sources: mini-swe-agent:agent.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:agent, mini-swe-agent:docker, mini-swe-agent:local, mini-swe-agent:retry

### RUN-06: Truthful status and operational statistics

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Full trajectories with per-call cost metadata, exit status and run configuration saved as JSON; a Textual inspector; batch progress summary; no traces, dashboards or aggregate statistics beyond batch output.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- traces: Trajectory with messages and model responses.
- usage_export: JSON trajectories; global model stats.
- project_stats: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- time_stats: Elapsed time in batch view.

- logs — Logs [supported]: Logging and trajectory files. Sources: mini-swe-agent:output.
- traces — Traces [partial]: Trajectory with messages and model responses. Sources: mini-swe-agent:output.
- failure_reasons — Failure reasons [supported]: Exit statuses and exceptions recorded. Sources: mini-swe-agent:swebench, mini-swe-agent:output.
- usage_export — Usage export [partial]: JSON trajectories; global model stats. Sources: mini-swe-agent:model, mini-swe-agent:output.
- task_stats — Task stats [supported]: Per-run cost and calls. Sources: mini-swe-agent:agent.
- agent_stats — Agent stats [not-applicable]: Single agent. Sources: mini-swe-agent:readme.
- project_stats — Project stats [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- time_stats — Time stats [partial]: Elapsed time in batch view. Sources: mini-swe-agent:swebench.
- failure_status — Failure status [supported]: Exit status table. Sources: mini-swe-agent:swebench.
- approval_status — Approval status [partial]: Confirmations in the REPL. Sources: mini-swe-agent:interactive.
- reconnect_consistency — Reconnect consistency [not-applicable]: No reconnect. Sources: mini-swe-agent:readme.
- performance_method — Performance method [not-applicable]: This catalog measures no speed; vendor claims of >74% SWE-bench Verified are not repeated as findings. Sources: mini-swe-agent:readme.
- responsiveness — Responsiveness [not-applicable]: None measured. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:output, mini-swe-agent:swebench, mini-swe-agent:model, mini-swe-agent:agent, mini-swe-agent:readme, mini-swe-agent:search, mini-swe-agent:interactive

### RUN-07: Per-task API and compute cost

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Per-call and cumulative model cost computed through litellm pricing (or provider-reported) and shown during runs and saved in the trajectory; cost limit enforces a stop; no compute cost, alerts or aggregation across runs beyond global stats.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- billed_vs_estimated: litellm cost estimates; ignore_errors mode when pricing missing.
- subscription_allowance: None.
- provider_reconciliation: None.
- compute_cost: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- task_cost — Task cost [supported]: Agent cost tracked per run. Sources: mini-swe-agent:agent, mini-swe-agent:model.
- session_cost — Session cost [supported]: Same. Sources: mini-swe-agent:agent.
- child_cost — Child cost [not-applicable]: No children. Sources: mini-swe-agent:readme.
- epic_cost — Epic cost [not-applicable]: None. Sources: mini-swe-agent:readme.
- harness_aggregation — Harness aggregation [not-applicable]: Single harness. Sources: mini-swe-agent:readme.
- billed_vs_estimated — Billed vs estimated [partial]: litellm cost estimates; ignore_errors mode when pricing missing. Sources: mini-swe-agent:model.
- tokens — Tokens [supported]: Token usage in responses saved in the trajectory. Sources: mini-swe-agent:output.
- subscription_allowance — Subscription allowance [unsupported]: None. Sources: mini-swe-agent:model.
- provider_reconciliation — Provider reconciliation [unsupported]: None. Sources: mini-swe-agent:model.
- compute_cost — Compute cost [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- refresh_delay — Refresh delay [supported]: Updated each model call. Sources: mini-swe-agent:model.
- alerts — Alerts [partial]: Cost limit stop only. Sources: mini-swe-agent:agent.
- export — Export [supported]: JSON trajectories. Sources: mini-swe-agent:output.
- drilldown — Drilldown [partial]: Per-message cost metadata. Sources: mini-swe-agent:model.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:agent, mini-swe-agent:model, mini-swe-agent:readme, mini-swe-agent:output, mini-swe-agent:search

### RUN-08: Compute-aware scheduling and verification

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Batch runs use a configurable worker count; no CPU or RAM awareness, queues with backpressure, or check deduplication.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- queues: Thread pool work list.
- cpu_ram: None.
- backpressure: None.
- host_awareness: None.

- concurrency — Concurrency [supported]: --workers. Sources: mini-swe-agent:swebench.
- queues — Queues [partial]: Thread pool work list. Sources: mini-swe-agent:swebench.
- cpu_ram — Cpu ram [unsupported]: None. Sources: mini-swe-agent:swebench.
- backpressure — Backpressure [unsupported]: None. Sources: mini-swe-agent:swebench.
- host_awareness — Host awareness [unsupported]: None. Sources: mini-swe-agent:swebench.
- check_reuse — Check reuse [unsupported]: None. Sources: mini-swe-agent:swebench.
- check_deduplication — Check deduplication [unsupported]: None. Sources: mini-swe-agent:swebench.
- bounded_tests — Bounded tests [unsupported]: None. Sources: mini-swe-agent:swebench.
- cancellation — Cancellation [partial]: Interrupt the batch process; futures cancelled. Sources: mini-swe-agent:swebench.
- fairness — Fairness [unsupported]: None. Sources: mini-swe-agent:swebench.
- recovery — Recovery [partial]: Finished instances skipped on rerun. Sources: mini-swe-agent:swebench.
- shared_limits — Shared limits [unsupported]: Limits are per agent. Sources: mini-swe-agent:agent.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench, mini-swe-agent:agent

## Human collaboration and control

### TEAM-01: Live human collaboration

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Single-user local tool: no sharing, comments or presence.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- visibility: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- discussion: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- comments: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- presence: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- visibility — Visibility [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- discussion — Discussion [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- comments — Comments [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- presence — Presence [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- coprompting — Coprompting [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- ordering — Ordering [not-applicable]: None. Sources: mini-swe-agent:readme.
- conflicts — Conflicts [not-applicable]: None. Sources: mini-swe-agent:readme.
- attribution — Attribution [not-applicable]: None. Sources: mini-swe-agent:readme.
- permissions — Permissions [not-applicable]: None. Sources: mini-swe-agent:readme.
- plan_restrictions — Plan restrictions [not-applicable]: None. Sources: mini-swe-agent:readme.
- owner_online — Owner online [not-applicable]: None. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:readme

### TEAM-02: Roles and access boundaries

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No roles, membership or SSO; local library.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- organization_roles: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- project_roles: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- task_roles: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- membership: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- organization_roles — Organization roles [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- project_roles — Project roles [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- task_roles — Task roles [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- membership — Membership [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- sso — Sso [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- provisioning — Provisioning [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- unauthorized_access — Unauthorized access [not-applicable]: No access model. Sources: mini-swe-agent:readme.
- revocation — Revocation [not-applicable]: None. Sources: mini-swe-agent:readme.
- plan_scope — Plan scope [not-applicable]: No plans. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:readme

### TEAM-03: Approvals and temporary access

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Confirm mode asks before each command unless it matches a whitelist regex; yolo mode runs immediately; environments provide containment; no network, secret or delivery approvals.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- network: None except container options.
- secret: None.
- delivery: Commands such as git push are ordinary commands subject to confirmation.
- revoke: Change mode with /c.

- command — Command [supported]: Confirm mode with whitelist_actions. Sources: mini-swe-agent:interactive.
- tool — Tool [not-applicable]: Bash is the only tool. Sources: mini-swe-agent:readme.
- network — Network [unsupported]: None except container options. Sources: mini-swe-agent:docker.
- secret — Secret [unsupported]: None. Sources: mini-swe-agent:interactive.
- delivery — Delivery [partial]: Commands such as git push are ordinary commands subject to confirmation. Sources: mini-swe-agent:interactive.
- approver_identity — Approver identity [not-applicable]: Single user. Sources: mini-swe-agent:readme.
- lifetime — Lifetime [not-applicable]: None. Sources: mini-swe-agent:readme.
- allow — Allow [supported]: Enter to confirm; whitelist. Sources: mini-swe-agent:interactive.
- deny — Deny [supported]: Reject with a message. Sources: mini-swe-agent:interactive.
- revoke — Revoke [partial]: Change mode with /c. Sources: mini-swe-agent:interactive.
- tunnels — Tunnels [not-applicable]: None. Sources: mini-swe-agent:readme.
- defaults — Defaults [supported]: confirm is default; yolo opt-in. Sources: mini-swe-agent:minidoc.
- safer_modes — Safer modes [supported]: Confirm and sandbox environments. Sources: mini-swe-agent:interactive, mini-swe-agent:envdoc.
- enforcement — Enforcement [supported]: Enforced in the interactive agent code. Sources: mini-swe-agent:interactive.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:interactive, mini-swe-agent:readme, mini-swe-agent:docker, mini-swe-agent:minidoc, mini-swe-agent:envdoc

### TEAM-04: Decision and activity history

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Trajectory files are the only history: full messages, config and metadata, locally stored and mutable; no attribution, approvals log or retention policy.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- approvals: Rejections appear in messages.
- edits: Only through commands in messages.
- decisions: None.
- cross_session_persistence: Last run file overwritten by default; custom output path possible.

- actions — Actions [supported]: Messages and actions in trajectory. Sources: mini-swe-agent:output.
- approvals — Approvals [partial]: Rejections appear in messages. Sources: mini-swe-agent:interactive.
- edits — Edits [partial]: Only through commands in messages. Sources: mini-swe-agent:output.
- discussion — Discussion [supported]: Full conversation. Sources: mini-swe-agent:output.
- decisions — Decisions [unsupported]: None. Sources: mini-swe-agent:output.
- cross_user_persistence — Cross user persistence [not-applicable]: Single user. Sources: mini-swe-agent:readme.
- cross_session_persistence — Cross session persistence [partial]: Last run file overwritten by default; custom output path possible. Sources: mini-swe-agent:mini.
- export — Export [supported]: JSON. Sources: mini-swe-agent:output.
- mutability — Mutability [supported]: Files are editable. Sources: mini-swe-agent:output.
- retention — Retention [unsupported]: None. Sources: mini-swe-agent:output.
- integrity — Integrity [unsupported]: None. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:output, mini-swe-agent:interactive, mini-swe-agent:readme, mini-swe-agent:mini

### TEAM-05: Review experience and guidance

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

The inspector browses trajectory steps; there is no diff review, comments or review queue.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- pending_reviews: None.
- change_understanding: Step-by-step trajectory view.
- comments: None.
- request_changes: None.

- pending_reviews — Pending reviews [unsupported]: None. Sources: mini-swe-agent:inspector.
- change_understanding — Change understanding [partial]: Step-by-step trajectory view. Sources: mini-swe-agent:inspector.
- evidence_navigation — Evidence navigation [supported]: Step navigation with keyboard. Sources: mini-swe-agent:inspector.
- comments — Comments [unsupported]: None. Sources: mini-swe-agent:inspector.
- request_changes — Request changes [unsupported]: None. Sources: mini-swe-agent:inspector.
- shipping_guidance — Shipping guidance [unsupported]: None. Sources: mini-swe-agent:inspector.
- effort — Effort [unsupported]: None. Sources: mini-swe-agent:inspector.
- missed_evidence — Missed evidence [unsupported]: None. Sources: mini-swe-agent:inspector.
- keyboard — Keyboard [supported]: Documented key bindings (step, scroll, palette). Sources: mini-swe-agent:inspector.
- narrow_screen — Narrow screen [partial]: Terminal UI. Sources: mini-swe-agent:inspector.
- context_switching — Context switching [partial]: Multiple trajectories 1/3 in inspector. Sources: mini-swe-agent:inspector.
- roles — Roles [unsupported]: None. Sources: mini-swe-agent:inspector.
- retained_context — Retained context [supported]: Full trajectory. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:inspector, mini-swe-agent:output

### TEAM-06: Colleague takeover and continuity

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No takeover: trajectories are files that another person can open, but runs cannot be handed over.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- ongoing_takeover: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- same_task: Trajectory file shareable manually.
- session_history: Same.
- decisions: In messages.

- ongoing_takeover — Ongoing takeover [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- same_task — Same task [partial]: Trajectory file shareable manually. Sources: mini-swe-agent:output.
- session_history — Session history [partial]: Same. Sources: mini-swe-agent:output.
- decisions — Decisions [partial]: In messages. Sources: mini-swe-agent:output.
- artifacts — Artifacts [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- permissions — Permissions [not-applicable]: None. Sources: mini-swe-agent:readme.
- runtime_context — Runtime context [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- owner_disconnect — Owner disconnect [unsupported]: Run ends. Sources: mini-swe-agent:agent.
- ownership — Ownership [not-applicable]: None. Sources: mini-swe-agent:readme.
- attribution — Attribution [not-applicable]: None. Sources: mini-swe-agent:readme.
- context_loss — Context loss [not-applicable]: None. Sources: mini-swe-agent:readme.
- harness_transfer — Harness transfer [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:output, mini-swe-agent:readme, mini-swe-agent:agent

## Verification and delivery

### SHIP-01: Retained artifacts and outputs

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Retained outputs are trajectories and, in benchmark mode, patches and result files; no gallery, screenshots or previews.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- diffs: Batch mode stores model patches; interactive mode has none.
- documents: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- screenshots: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- recordings: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- diffs — Diffs [partial]: Batch mode stores model patches; interactive mode has none. Sources: mini-swe-agent:swebench.
- documents — Documents [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- screenshots — Screenshots [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- recordings — Recordings [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- previews — Previews [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- test_outputs — Test outputs [partial]: Command outputs inside the trajectory. Sources: mini-swe-agent:output.
- reviewer_outputs — Reviewer outputs [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- gallery — Gallery [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- task_links — Task links [partial]: Instance ID to trajectory file. Sources: mini-swe-agent:swebench.
- revision_links — Revision links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- persistence — Persistence [supported]: JSON files. Sources: mini-swe-agent:output.
- child_findings — Child findings [not-applicable]: None. Sources: mini-swe-agent:readme.
- prompt_leakage — Prompt leakage [not-applicable]: None. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench, mini-swe-agent:search, mini-swe-agent:output, mini-swe-agent:readme

### SHIP-02: Verification and spec gates

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No gates; checks are whatever commands the model runs, and benchmark runners evaluate patches outside the agent.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- tests: Model-run commands.
- lint: Same.
- types: Same.
- ci: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- tests — Tests [partial]: Model-run commands. Sources: mini-swe-agent:readme.
- lint — Lint [partial]: Same. Sources: mini-swe-agent:readme.
- types — Types [partial]: Same. Sources: mini-swe-agent:readme.
- ci — Ci [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- independent_review — Independent review [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- spec_checks — Spec checks [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- required_checks — Required checks [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- failure_blocking — Failure blocking [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- requirement_conflicts — Requirement conflicts [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- configured_policy — Configured policy [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- enforced_policy — Enforced policy [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:search

### SHIP-03: Git and pull request workflow

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No Git integration: the model runs git and gh through bash if instructed.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- branches: Model-run git.
- worktrees: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- commits: Model-run git.
- pull_requests: Model-run gh; README says to tell the LM to open PRs.

- branches — Branches [partial]: Model-run git. Sources: mini-swe-agent:readme.
- worktrees — Worktrees [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- commits — Commits [partial]: Model-run git. Sources: mini-swe-agent:readme.
- pull_requests — Pull requests [partial]: Model-run gh; README says to tell the LM to open PRs. Sources: mini-swe-agent:readme.
- comments — Comments [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- ci_response — Ci response [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- rebase — Rebase [partial]: Model-run git. Sources: mini-swe-agent:readme.
- conflicts — Conflicts [partial]: Model-run git. Sources: mini-swe-agent:readme.
- product_state — Product state [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:search

### SHIP-04: Merge and deployment authority

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No merge, release or deploy features.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- approve: Command confirmation only.
- merge: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- release: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- deploy: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- approve — Approve [partial]: Command confirmation only. Sources: mini-swe-agent:interactive.
- merge — Merge [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- release — Release [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- deploy — Deploy [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- exact_revision — Exact revision [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- authority_enforcement — Authority enforcement [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:interactive, mini-swe-agent:search

### SHIP-05: Completion evidence and rollback

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No delivery tracking, rollback or evidence links.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- commit_links: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- pr_links: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- build_links: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- deployment_links: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- commit_links — Commit links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- pr_links — Pr links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- build_links — Build links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- deployment_links — Deployment links [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- delivery_status — Delivery status [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- retained_evidence — Retained evidence [partial]: Trajectory. Sources: mini-swe-agent:output.
- rollback — Rollback [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- revert — Revert [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- stale_approval_invalidation — Stale approval invalidation [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:search, mini-swe-agent:output

### SHIP-06: Batch review and high-volume shipping

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

No batch review or shipping queue; batch mode evaluates patches against benchmarks.

Availability: generally-available
Delivery: native
Scope: surfaces=mini-extra batch runners and inspector (SWE-bench, ProgramBench); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- review_queue: None.
- summaries: Exit-status table.
- grouping: None.
- risk_size_filters: None.

- review_queue — Review queue [unsupported]: None. Sources: mini-swe-agent:swebench.
- summaries — Summaries [partial]: Exit-status table. Sources: mini-swe-agent:swebench.
- grouping — Grouping [unsupported]: None. Sources: mini-swe-agent:swebench.
- risk_size_filters — Risk size filters [unsupported]: None. Sources: mini-swe-agent:swebench.
- findings — Findings [unsupported]: None. Sources: mini-swe-agent:swebench.
- artifacts — Artifacts [unsupported]: None. Sources: mini-swe-agent:swebench.
- request_changes — Request changes [unsupported]: None. Sources: mini-swe-agent:swebench.
- exact_revision_approval — Exact revision approval [unsupported]: None. Sources: mini-swe-agent:swebench.
- dependency_queues — Dependency queues [unsupported]: None. Sources: mini-swe-agent:swebench.
- next_actions — Next actions [unsupported]: None. Sources: mini-swe-agent:swebench.
- stale_approval_invalidation — Stale approval invalidation [unsupported]: None. Sources: mini-swe-agent:swebench.
- check_deduplication — Check deduplication [unsupported]: None. Sources: mini-swe-agent:swebench.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:swebench

## Ownership, privacy and economics

### OWN-01: Source and license

Verification status: Agent cross-checked (agent-cross-checked)
Support: supported

Single public repository under the MIT license at v2.4.6 covering the agent, environments, models and runners; no closed or hosted component.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL), mini-extra batch runners and inspector (SWE-bench, ProgramBench), Python library (agent, model and environment classes); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- client_license — Client license [supported]: MIT. Sources: mini-swe-agent:license, mini-swe-agent:repo.
- server_license — Server license [not-applicable]: No server. Sources: mini-swe-agent:readme.
- hosted_extensions — Hosted extensions [not-applicable]: No hosted service. Sources: mini-swe-agent:readme.
- source_visibility — Source visibility [supported]: Public. Sources: mini-swe-agent:repo.
- source_revision — Source revision [supported]: Tag v2.4.6. Sources: mini-swe-agent:rel.
- hosted_parity — Hosted parity [not-applicable]: No hosted service. Sources: mini-swe-agent:readme.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:license, mini-swe-agent:repo, mini-swe-agent:readme, mini-swe-agent:rel

### OWN-02: Hosting and exit

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Fully local library; all data in local .env and trajectory files with JSON export; no vendor account; no import.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- vendor_dependencies: Model providers only.
- import: Inspector reads trajectory files.

- self_hosting — Self hosting [supported]: Runs entirely on user machine. Sources: mini-swe-agent:readme.
- vendor_dependencies — Vendor dependencies [partial]: Model providers only. Sources: mini-swe-agent:modelsdoc.
- export — Export [supported]: JSON trajectories. Sources: mini-swe-agent:output.
- import — Import [partial]: Inspector reads trajectory files. Sources: mini-swe-agent:inspector.
- formats — Formats [supported]: JSON and YAML. Sources: mini-swe-agent:output.
- cancellation — Cancellation [not-applicable]: No account. Sources: mini-swe-agent:readme.
- account_independence — Account independence [supported]: No account. Sources: mini-swe-agent:readme.
- history_portability — History portability [supported]: Plain files. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:modelsdoc, mini-swe-agent:output, mini-swe-agent:inspector

### OWN-03: Data handling and telemetry

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Code and prompts go to the model provider you configure through litellm; keys stay in local .env; trajectories are local; no telemetry code was found in src or dependencies (searched telemetry, analytics, posthog, sentry); no consent dialog exists.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- training: Provider policy not captured.
- residency: Provider dependent.
- subprocessors: Provider dependent.
- telemetry_disclosure: No telemetry code or statement found (rg for telemetry, analytics, posthog and sentry: 0 hits in src, docs, tests).

- code_data — Code data [supported]: To the configured provider. Sources: mini-swe-agent:modelsdoc.
- prompts — Prompts [supported]: Same. Sources: mini-swe-agent:modelsdoc.
- credentials — Credentials [supported]: Local .env file. Sources: mini-swe-agent:globalconfig.
- logs — Logs [supported]: Local logging and trajectories. Sources: mini-swe-agent:output.
- artifacts — Artifacts [supported]: Local. Sources: mini-swe-agent:output.
- retention — Retention [supported]: Files persist until deleted. Sources: mini-swe-agent:output.
- deletion — Deletion [supported]: Delete files. Sources: mini-swe-agent:output.
- training — Training [unknown]: Provider policy not captured.
- residency — Residency [unknown]: Provider dependent.
- subprocessors — Subprocessors [unknown]: Provider dependent.
- telemetry_disclosure — Telemetry disclosure [partial]: No telemetry code or statement found (rg for telemetry, analytics, posthog and sentry: 0 hits in src, docs, tests). Sources: mini-swe-agent:search.
- telemetry_defaults — Telemetry defaults [supported]: None found in src (bounded search). Sources: mini-swe-agent:search.
- telemetry_controls — Telemetry controls [not-applicable]: Nothing to control. Sources: mini-swe-agent:search.
- consent_prompt — Consent prompt [not-applicable]: None needed. Sources: mini-swe-agent:search.
- provider_separation — Provider separation [supported]: Providers are user chosen. Sources: mini-swe-agent:modelsdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:modelsdoc, mini-swe-agent:globalconfig, mini-swe-agent:output, mini-swe-agent:search

### OWN-04: Price and included usage

Verification status: Agent cross-checked (agent-cross-checked)
Support: supported

Free MIT software with no paid plan; model usage is billed by providers.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- currency — Currency [not-applicable]: No ADE price. Sources: mini-swe-agent:readme.
- interval — Interval [not-applicable]: None. Sources: mini-swe-agent:readme.
- seats — Seats [not-applicable]: None. Sources: mini-swe-agent:readme.
- seat_minimum — Seat minimum [not-applicable]: None. Sources: mini-swe-agent:readme.
- workspace_fees — Workspace fees [not-applicable]: None. Sources: mini-swe-agent:readme.
- included_usage — Included usage [not-applicable]: None. Sources: mini-swe-agent:license.
- credits — Credits [not-applicable]: None. Sources: mini-swe-agent:readme.
- trial — Trial [not-applicable]: None. Sources: mini-swe-agent:readme.
- overages — Overages [not-applicable]: None. Sources: mini-swe-agent:readme.
- taxes — Taxes [not-applicable]: None. Sources: mini-swe-agent:readme.
- model_cost — Model cost [supported]: Provider costs separate; cost tracked and limited by cost_limit. Sources: mini-swe-agent:agent, mini-swe-agent:readme.
- compute_cost — Compute cost [supported]: User pays containers or Modal. Sources: mini-swe-agent:envdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:license, mini-swe-agent:agent, mini-swe-agent:envdoc

### OWN-05: Scenario cost and controls

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Software cost is zero for any team size; model and compute costs depend on provider and use and are unpublished.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- task_assumptions: None.
- concurrency: None.
- tokens: None.

- one_person — One person [supported]: $0 software. Sources: mini-swe-agent:license.
- two_people — Two people [supported]: $0. Sources: mini-swe-agent:license.
- five_people — Five people [supported]: $0. Sources: mini-swe-agent:license.
- task_assumptions — Task assumptions [unknown]: None.
- concurrency — Concurrency [unknown]: None.
- tokens — Tokens [unknown]: None.
- compute_hours — Compute hours [not-applicable]: None. Sources: mini-swe-agent:readme.
- fixed_cost — Fixed cost [supported]: Zero. Sources: mini-swe-agent:license.
- variable_cost — Variable cost [supported]: Model usage. Sources: mini-swe-agent:agent.
- included_cost — Included cost [not-applicable]: None. Sources: mini-swe-agent:readme.
- unknown_cost — Unknown cost [supported]: Model spend unknown without workload. Sources: mini-swe-agent:agent.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:license, mini-swe-agent:readme, mini-swe-agent:agent

### OWN-06: Availability and support

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Maintained research-lab open-source project (Princeton and Stanford SWE-bench team): v2.4.6 on 2026-07-23, default branch active to 2026-11-09, Slack and GitHub issues for support; no response commitments.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- response_commitments: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).

- release_stage — Release stage [supported]: Versioned releases v2.x. Sources: mini-swe-agent:rel.
- access_restrictions — Access restrictions [supported]: Public. Sources: mini-swe-agent:readme.
- support — Support [supported]: Slack and GitHub issues. Sources: mini-swe-agent:readme.
- response_commitments — Response commitments [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- release_activity — Release activity [supported]: Releases July 2026; commits to November 2026. Sources: mini-swe-agent:rel, mini-swe-agent:repo.
- status_history — Status history [not-applicable]: No hosted service. Sources: mini-swe-agent:readme.
- deprecation — Deprecation [supported]: v1 branch and v2 migration guide. Sources: mini-swe-agent:rel.
- export_options — Export options [supported]: JSON. Sources: mini-swe-agent:output.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:rel, mini-swe-agent:readme, mini-swe-agent:search, mini-swe-agent:repo, mini-swe-agent:output

### OWN-07: Enterprise assurance

Verification status: Agent cross-checked (agent-cross-checked)
Support: supported

No ISO or SOC 2 claims; the security policy lists a vulnerability contact only. The project is open-source software with no hosted service to certify.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- iso_scope — Iso scope [not-applicable]: No hosted service and no claim. Sources: mini-swe-agent:security.
- iso_issuer — Iso issuer [not-applicable]: None. Sources: mini-swe-agent:security.
- iso_validity — Iso validity [not-applicable]: None. Sources: mini-swe-agent:security.
- soc2_type — Soc2 type [not-applicable]: None. Sources: mini-swe-agent:security.
- soc2_scope — Soc2 scope [not-applicable]: None. Sources: mini-swe-agent:security.
- soc2_period — Soc2 period [not-applicable]: None. Sources: mini-swe-agent:security.
- report_access — Report access [not-applicable]: None. Sources: mini-swe-agent:security.
- verification_date — Verification date [not-applicable]: None. Sources: mini-swe-agent:security.
- evidence_status — Evidence status [supported]: No assurance claim exists in first-party sources. Sources: mini-swe-agent:security.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:security

### OWN-08: Useful free tier and evaluation access

Verification status: Agent cross-checked (agent-cross-checked)
Support: partial

Free open-source software; costs are model usage and any compute you add; no trial or paid tier.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

Limitations:
- paid_only: Not paid.
- early_access: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence).
- free_scenarios: Single-user local use; no collaboration, remote or mobile features.

- ongoing_free — Ongoing free [supported]: Free MIT download. Sources: mini-swe-agent:license.
- trial — Trial [not-applicable]: None. Sources: mini-swe-agent:readme.
- credits_trial — Credits trial [not-applicable]: None. Sources: mini-swe-agent:readme.
- self_host_cost — Self host cost [supported]: User pays models and compute. Sources: mini-swe-agent:readme.
- paid_only — Paid only [unsupported]: Not paid. Sources: mini-swe-agent:search.
- early_access — Early access [unsupported]: Not found in the bounded search of src, docs and tests (terms and counts in the search evidence). Sources: mini-swe-agent:search.
- free_scenarios — Free scenarios [partial]: Single-user local use; no collaboration, remote or mobile features. Sources: mini-swe-agent:readme.
- card_required — Card required [supported]: No card for the software. Sources: mini-swe-agent:readme.
- usage_caps — Usage caps [supported]: Only the cost limit you set. Sources: mini-swe-agent:agent.
- paid_evaluation — Paid evaluation [supported]: Evaluation needs only a model key or local model. Sources: mini-swe-agent:modelsdoc.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:license, mini-swe-agent:readme, mini-swe-agent:search, mini-swe-agent:agent, mini-swe-agent:modelsdoc

### OWN-09: Pricing transparency

Verification status: Agent cross-checked (agent-cross-checked)
Support: supported

Nothing is sold; pricing transparency concerns only third-party model providers.

Availability: generally-available
Delivery: native
Scope: surfaces=mini interactive command line (REPL); version=v2.4.6 (commit a83fcae82d2a08f0ee0c688f9d137b3566c097f8, 2026-07-22); default branch checked 2026-11-16; plan=not specified; os=not specified; execution_modes=not specified; region=not specified; harness=not specified.

- units — Units [not-applicable]: None. Sources: mini-swe-agent:readme.
- included_limits — Included limits [not-applicable]: None. Sources: mini-swe-agent:readme.
- overages — Overages [not-applicable]: None. Sources: mini-swe-agent:readme.
- compute_markup — Compute markup [not-applicable]: None. Sources: mini-swe-agent:readme.
- model_markup — Model markup [not-applicable]: None; direct provider billing. Sources: mini-swe-agent:readme.
- minimum_commitment — Minimum commitment [not-applicable]: None. Sources: mini-swe-agent:readme.
- cancellation — Cancellation [not-applicable]: None. Sources: mini-swe-agent:readme.
- sales_only_unknowns — Sales only unknowns [not-applicable]: None. Sources: mini-swe-agent:readme.
- transparency_classification — Transparency classification [supported]: No price; free software. Sources: mini-swe-agent:license.

Checked 2026-11-16 (6 days old on 2026-11-22); agent cross-checked 2026-11-17.

Sources: mini-swe-agent:readme, mini-swe-agent:license

## Sources

### mini-swe-agent:interactive

InteractiveAgent modes. Three modes: confirm (default, user confirms each command unless whitelisted by regex), yolo, human takeover; Ctrl+C interrupts; /c /y /u switch modes.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/agents/interactive.py#L24-L30
- Locator: src/minisweagent/agents/interactive.py InteractiveAgentConfig mode human/confirm/yolo, whitelist_actions, _ask_confirmation_or_interrupt
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/agents/interactive.py

### mini-swe-agent:mini

mini run script. mini CLI with model, task, yolo, cost limit, config and trajectory output (last_mini_run.traj.json in the global config directory).

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/run/mini.py#L54-L65
- Locator: src/minisweagent/run/mini.py options -m -t -y -l -c -o and DEFAULT_OUTPUT_FILE
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/run/mini.py

### mini-swe-agent:envdoc

Environment classes docs. local runs with subprocess.run and no isolation; docker exec; Singularity/Apptainer; SWE-ReX Docker and Modal cloud; bubblewrap (Linux only) unprivileged sandbox; ConTree.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/advanced/environments.md
- Locator: environment list
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:pyproject

Package metadata. Python 3.10+, OS independent; dependencies litellm, textual, typer, rich; scripts mini, mini-swe-agent, mini-extra.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/pyproject.toml#L33-L45
- Locator: pyproject.toml dependencies, scripts, classifiers
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: pyproject.toml

### mini-swe-agent:output

Output files. Full run history, config and metadata saved as .traj.json.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/usage/output_files.md
- Locator: Trajectory files
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:minidoc

mini usage docs. REPL-style CLI for local use; confirm/yolo/human modes; history of the last run saved to the global config directory.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/usage/mini.md
- Locator: Command line options; Modes of operation
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:search

Bounded search of the pinned release. Search terms and hit counts (src/docs/tests): mcp 0/0/0; worktree 0/0/0; schedule|cron 0/0/0; webhook 0/0/0; skill 0/0/0; AGENTS.md|CLAUDE.md|.cursorrules 0/0/0; telemetry|analytics|posthog|sentry 0/0/0; snapshot|checkpoint|resume 0/0/0; screenshot 0/0/0; oauth|subscription|login 2/0/0 (docker.py:41 login-shell flag, swebench.yaml:125 BASH_ENV, neither a product login); subagent|delegate|spawn 2/0/1 (programbench.yaml prompt text and a timeout test only); github issue|gh pr 3/5/1 (error-message links and docs links only).

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent
- Locator: rg -ci over src/, docs/ and tests/ at v2.4.6
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent

### mini-swe-agent:readme

mini-SWE-agent README. Minimal agent: bash is the only tool, linear message history, subprocess.run per action; local, docker/podman, singularity, bubblewrap, contree environments; models via litellm, openrouter, portkey, requesty; >74% SWE-bench Verified; MIT.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/README.md
- Locator: README.md intro, motivation, deployable and compatible bullets
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:agent

DefaultAgent and AgentConfig. About 190-line agent loop; stops at step, cost or wall-time limits (cost limit default $3); trajectory saved to output_path; no tools besides bash.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/agents/default.py#L19-L36
- Locator: src/minisweagent/agents/default.py AgentConfig (step_limit, cost_limit=3.0, wall_time_limit_seconds, max_consecutive_format_errors) and run/query/execute_actions
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/agents/default.py

### mini-swe-agent:modelsdoc

Model setup quickstart. API keys through mini-extra config setup, env vars or .env; litellm supports most providers, plus OpenRouter and Portkey.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/models/quickstart.md
- Locator: Setting API keys
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:globalconfig

Global configuration. API keys and defaults stored in a .env file via mini-extra config; env vars override.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/advanced/global_configuration.md
- Locator: Setting global configuration
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:extra

mini-extra subcommands. config, inspect, swebench, swebench-single and programbench subcommands.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/run/utilities/mini_extra.py#L13-L19
- Locator: src/minisweagent/run/utilities/mini_extra.py subcommands list
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/run/utilities/mini_extra.py

### mini-swe-agent:rel

mini-swe-agent v2.4.6 release. Latest tagged release at pass time.

- URL: https://github.com/SWE-agent/mini-swe-agent/releases/tag/v2.4.6
- Locator: Release 2026-07-23
- Evidence kind: documented
- Source date: 2026-07-23
- Retrieved: 2026-11-16

### mini-swe-agent:inspector

Inspector docs. Textual-based trajectory browser for .traj.json files (mini-extra inspect).

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/usage/inspector.md
- Locator: Overview
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16

### mini-swe-agent:swebench

SWE-bench batch runner. Batch mode runs many instances in parallel worker threads with a progress display.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/run/benchmarks/swebench.py#L209
- Locator: src/minisweagent/run/benchmarks/swebench.py --workers, concurrent.futures
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/run/benchmarks/swebench.py

### mini-swe-agent:model

LitellmModel cost tracking. Every model call records cost and adds it to global stats; cost_tracking can ignore missing cost info.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/models/litellm_model.py#L36
- Locator: src/minisweagent/models/litellm_model.py cost_tracking, GLOBAL_MODEL_STATS
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/models/litellm_model.py

### mini-swe-agent:local

LocalEnvironment. Executes bash commands on the host with a 30 second default per-command timeout.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/environments/local.py#L13-L16
- Locator: src/minisweagent/environments/local.py LocalEnvironmentConfig (cwd, env, timeout=30)
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/environments/local.py

### mini-swe-agent:docker

DockerEnvironment. Runs each action with docker exec in a container started per run; executable can be podman; container lifetime 2h default.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/environments/docker.py#L15-L36
- Locator: src/minisweagent/environments/docker.py DockerEnvironmentConfig (executable, forward_env, timeout, container_timeout 2h)
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/environments/docker.py

### mini-swe-agent:bwrap

BubblewrapEnvironment. Linux-only unprivileged bubblewrap sandbox for commands.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/environments/extra/bubblewrap.py#L10
- Locator: src/minisweagent/environments/extra/bubblewrap.py
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/environments/extra/bubblewrap.py

### mini-swe-agent:retry

Model retry utility. Model queries retry with exponential wait and a bounded attempt count via tenacity.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/src/minisweagent/models/utils/retry.py#L9
- Locator: src/minisweagent/models/utils/retry.py tenacity Retrying
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: src/minisweagent/models/utils/retry.py

### mini-swe-agent:license

MIT license. MIT license, copyright Kilian A. Lieret and Carlos E. Jimenez.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/LICENSE.md
- Locator: LICENSE.md
- Evidence kind: source-inspected
- Source date: Not stated
- Retrieved: 2026-11-16
- Commit: a83fcae82d2a08f0ee0c688f9d137b3566c097f8
- Path: LICENSE.md

### mini-swe-agent:repo

GitHub repository metadata. 9,661 stars on 2026-11-16; MIT; not archived; last pushed 2026-11-09.

- URL: https://api.github.com/repos/SWE-agent/mini-swe-agent
- Locator: stargazers_count, license, pushed_at
- Evidence kind: documented
- Source date: 2026-11-16
- Retrieved: 2026-11-16

### mini-swe-agent:security

Security policy. Only a vulnerability contact list; no certifications.

- URL: https://github.com/SWE-agent/mini-swe-agent/blob/v2.4.6/docs/SECURITY.md
- Locator: Reporting a Vulnerability
- Evidence kind: documented
- Source date: Not stated
- Retrieved: 2026-11-16
