39 of 51 criteria have a sourced finding; 36 were also cross-checked by a second agent. Every comparison that includes OpenHands links here for scope, limitations, atomic answers and sources.
Humans approve the rubric, templates and prototype; they do not review every finding. An agent source cross-check is not a hands-on product test. A missing sourced finding is not evidence that a capability is absent.
Canvas provides desktop installation paths, browser UI and documented phone/tablet browser access. A separate public source-built native mobile MVP supports iOS/Android direct Agent Server access, including conversation listing and message sending; Cloud login and Canvas pairing remain planned. The product-owned agent-canvas command launches the local stack.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, Public source-built OpenHands native mobile MVP for direct Agent Server access · Version: e14b76cca01a71597cee242c202eed730e745ce3 Canvas launcher; source-built native mobile MVP 47f815b1d111b6a71b405dcb77acb3682da59b85 · Plan: Canvas desktop/browser and launcher; native mobile is limited to direct Agent Server access, without Cloud login or Canvas pairing. · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
Not separately cross-checked
Only the named atomic answers are established by this source pass. Remaining atomic answers retain their documented gaps.
9 atomic answers
Desktop os [partial] macOS, Windows and Linux installation is advertised; minimum OS versions unverified. Sources: openhands:nov07-canvas.
Mobile apps [partial] The native Expo source MVP builds for iOS and Android and can list/stream/send to a reachable Agent Server using a session API key. Requires native toolchains; Cloud login, QR pairing and laptop facade are not built. Sources: openhands:gap39-mobile-doc, openhands:gap39-mobile-code.
Mobile browser [supported] Open Canvas in a phone/tablet browser through Tailscale or authenticated ngrok/public mode. Sources: openhands:gap-phone.
Cli [supported] The agent-canvas command launches the local server/automation/frontend stack and supports help/info/version and deployment flags. This is launcher scope. Sources: openhands:gap39-cli.
Editor extensions [unknown] Editor extensions: The November first-party documentation and repository pass did not establish client access at the rubric's required scope. No installed-product fixture was run.
Task actions [unknown] Task actions: The November first-party documentation and repository pass did not establish client access at the rubric's required scope. No installed-product fixture was run.
Notifications [unknown] Notifications: The November first-party documentation and repository pass did not establish client access at the rubric's required scope. No installed-product fixture was run.
Cross device continuity [partial] Phone/tablet browsers can reach the same host Canvas/backend through Tailscale or authenticated ngrok; continuing a specific task across devices is not described and no desktop-phone handoff fixture was run. Sources: openhands:gap-phone.
Canvas integrates OpenHands and ACP agents with custom stdio ACP commands.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
Model providers [supported] OpenHands agent: any litellm-supported model with your own key, custom model and base URL; saved LLM profiles switch models mid-conversation. ACP agents use their own model settings. Sources: openhands:b09-llm, openhands:b09-readme.
Custom adapters [supported] Custom preset accepts a stdio ACP launch command and optional custom model ID. Sources: openhands:gap-acp.
Session integration [supported] ACP integration for compatible harnesses. Sources: openhands:nov07-canvas.
Mixed harness operation [unknown] Mixed harness operation: The November first-party documentation and repository pass did not establish agent and model choice at the rubric's required scope. No installed-product fixture was run.
Local ACP providers can reuse native logins, while clean containers/cloud require credentials and have provider limits.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
7 atomic answers
Agent subscriptions [supported] Existing Claude Code, Codex or Gemini subscription access. Sources: openhands:nov07-canvas.
Api accounts [supported] OpenHands agent can use an LLM key. Sources: openhands:nov07-canvas.
Vendor credits [unknown] Vendor credits: The November first-party documentation and repository pass did not establish subscriptions, api keys and vendor credits at the rubric's required scope. No installed-product fixture was run.
Local models [unknown] Local models: The November first-party documentation and repository pass did not establish subscriptions, api keys and vendor credits at the rubric's required scope. No installed-product fixture was run.
Account switching [unknown] Account switching: The November first-party documentation and repository pass did not establish subscriptions, api keys and vendor credits at the rubric's required scope. No installed-product fixture was run.
Vendor account requirement [unknown] Vendor account requirement: The November first-party documentation and repository pass did not establish subscriptions, api keys and vendor credits at the rubric's required scope. No installed-product fixture was run.
Provider restrictions [supported] Pi and OpenCode offered only for local/self-hosted backends, not OpenHands Cloud; same-container HOME can race on provider config. Sources: openhands:gap-acp.
No git [unknown] No git: The November first-party documentation and repository pass did not establish onboarding and project prerequisites at the rubric's required scope. No installed-product fixture was run.
Local git without remote [unknown] Local git without remote: The November first-party documentation and repository pass did not establish onboarding and project prerequisites at the rubric's required scope. No installed-product fixture was run.
Hosted git [unknown] Hosted git: The November first-party documentation and repository pass did not establish onboarding and project prerequisites at the rubric's required scope. No installed-product fixture was run.
Managed cloud setup [unknown] Managed cloud setup: The November first-party documentation and repository pass did not establish onboarding and project prerequisites at the rubric's required scope. No installed-product fixture was run.
Canvas uses the Agent Server REST API; SDK ownership is separate.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
8 atomic answers
Cli [partial] The agent-canvas CLI starts the full stack or a frontend/backend split; it has no commands to start or steer conversations. Sources: openhands:b09-cli.
Sdk [supported] Python SDK and TypeScript client live in separate OpenHands repositories. Sources: openhands:nov07-readme.
Mcp server [unknown] Mcp server: The November first-party documentation and repository pass did not establish programmable access at the rubric's required scope. No installed-product fixture was run.
Webhooks [supported] Inbound: automations fire on webhook events, including custom webhook sources with a URL and signing secret. Sources: openhands:b09-automations, openhands:b09-readme.
Authentication [unknown] Authentication: The November first-party documentation and repository pass did not establish programmable access at the rubric's required scope. No installed-product fixture was run.
Stability [unknown] Stability: The November first-party documentation and repository pass did not establish programmable access at the rubric's required scope. No installed-product fixture was run.
Plugins [supported] Customize includes a Plugins page on local backends; on an OpenHands Cloud backend the local Plugins page is hidden and Skills and Plugins opens Cloud settings. Sources: openhands:gap-customize.
Custom tools [unknown] Custom tools: The November first-party documentation and repository pass did not establish external tools and extensions at the rubric's required scope. No installed-product fixture was run.
Authentication [supported] MCP health distinguishes credential rejection from connection failure; reachable public MCP does not imply credential verification. Sources: openhands:gap-customize.
Permission scope [supported] Skills, MCP servers and secrets are scoped to active backend, not automatically copied across backends. Sources: openhands:gap-customize.
Integration owner [partial] Canvas ships the connection library; which component owns external calls was not established. Sources: openhands:nov07-canvas.
Surfaces: Agent Canvas · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
4 atomic answers
Desktop implementation [unknown] Desktop implementation: The November first-party documentation and repository pass did not establish client implementation at the rubric's required scope. No installed-product fixture was run.
Mobile implementation [unknown] Mobile implementation: The November first-party documentation and repository pass did not establish client implementation at the rubric's required scope. No installed-product fixture was run.
Issue-to-PR automations read tracker issues; Slack, GitHub and webhook events trigger agent runs. Other intake and sync atoms remain unverified.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: Earlier evidence keeps its cited revisions. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Agent Canvas issue-to-PR, Slack and GitHub/webhook automations with a configured backend and required tracker credentials, tools and secrets. Event/automation availability depends on backend support; these findings do not establish Cloud or source-built mobile parity. · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
9 atomic answers
Local creation [unknown] Local creation: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
Prompt intake [unknown] Prompt intake: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
External import [supported] Issue-to-PR automations read the title, description, acceptance criteria and linked dependencies of labeled GitHub, GitLab, Jira or Linear issues. Sources: openhands:b09-issue-pr.
Authoritative tracker [unknown] Authoritative tracker: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
Stable ids [unknown] Stable ids: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
Sync [unknown] Sync: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
Supported edits [unknown] Supported edits: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
External roundtrips [unknown] External roundtrips: The November first-party documentation and repository pass did not establish intake and external tracker relationship at the rubric's required scope. No installed-product fixture was run.
Canvas documents skills and a persistent workspace/user memory setting that can be enabled or disabled.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
10 atomic answers
Instructions [supported] Customize > Skills teaches domain instructions; /add-skill installs for a new conversation. Sources: openhands:gap-customize.
Skills [supported] Skills load when a conversation starts. Sources: openhands:gap-customize.
Project docs [unknown] Project docs: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
Memory [supported] Enabled persistent memory loads workspace and user memory into new OpenHands and ACP conversations. Sources: openhands:gap-customize.
Scope [supported] Saved configuration follows the selected backend; memory locations are workspace/user scoped. Sources: openhands:gap-customize.
Provenance [unknown] Provenance: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
Correction [unknown] Correction: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
Deletion [unknown] Deletion: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
Freshness [unknown] Freshness: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
Takeover continuity [unknown] Takeover continuity: The November first-party documentation and repository pass did not establish shared context and memory at the rubric's required scope. No installed-product fixture was run.
The Canvas Task List renders the latest TaskTrackerObservation from the current conversation, with position-derived IDs and todo/in-progress/done states. Bounded inspection found no independent native issue/backlog work-item store; saved automations are trigger configurations.
Scope
Surfaces: Agent Canvas · Version: e14b76cca01a71597cee242c202eed730e745ce3 native Canvas Task List and application stores/routes. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-13 (9 days old)
Agent cross-check
Not separately cross-checked
Only the named atomic answers are established by this source pass. Remaining atomic answers retain their documented gaps.
17 atomic answers
Create [unsupported] The inspected native interface renders agent TaskTracker observations rather than providing independent issue/backlog CRUD. Sources: openhands:gap39-task-list.
Edit [unknown] Edit: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Discuss [unknown] Discuss: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Prioritize [unknown] Prioritize: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Label [unknown] Label: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Assign [unknown] Assign: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Search [unknown] Search: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Filter [unknown] Filter: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Archive [unknown] Archive: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Reopen [unknown] Reopen: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Project epic views [unknown] Project epic views: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Idea parking [unknown] Idea parking: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Explicit dispatch [unknown] Explicit dispatch: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Task session separation [unsupported] Task rows derive directly from the current conversation event store and do not have independent stable issue identity. Sources: openhands:gap39-task-list.
Metadata [unknown] Metadata: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Durable states [partial] Task statuses are retained as conversation observations, not as separate persistent backlog work items. Sources: openhands:gap39-task-list.
External tracker requirement [unknown] External tracker requirement: The November first-party documentation and repository pass did not establish native issue management and idea backlog at the rubric's required scope. No installed-product fixture was run.
Concurrent tasks [unknown] Concurrent tasks: The November first-party documentation and repository pass did not establish parallel work at the rubric's required scope. No installed-product fixture was run.
Quotas [unknown] Quotas: The November first-party documentation and repository pass did not establish parallel work at the rubric's required scope. No installed-product fixture was run.
Observed concurrency [unknown] Observed concurrency: The November first-party documentation and repository pass did not establish parallel work at the rubric's required scope. No installed-product fixture was run.
Status visibility [partial] Canvas conversation UI; status correctness untested. Sources: openhands:nov07-readme.
Canvas launches linked child conversations and returns their results to the parent.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
Model harness effort [unknown] Model harness effort: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Messages [unknown] Messages: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Results return [supported] Child result returns to the parent conversation. Sources: openhands:gap-conversations.
Cross harness [unknown] Cross harness: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Authentication [unknown] Authentication: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Live progress [partial] Each working conversation shows its current stage as a live group in the chat; no parent-level live view of child progress is documented. Sources: openhands:b09-conversations.
Retained results [supported] The child conversation stays linked to its parent and its result is returned to the parent conversation. Sources: openhands:b09-conversations.
Followup [unknown] Followup: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Invocation [unknown] Invocation: The November first-party documentation and repository pass did not establish inspectable delegation and communication at the rubric's required scope. No installed-product fixture was run.
Automation Server owns schedules/events and Canvas exposes lifecycle/run history.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
Recurring tasks [unknown] Recurring tasks: The November first-party documentation and repository pass did not establish unattended triggers at the rubric's required scope. No installed-product fixture was run.
Completion conditions [unknown] Completion conditions: The November first-party documentation and repository pass did not establish unattended triggers at the rubric's required scope. No installed-product fixture was run.
Overlaps [supported] Concurrency-limit exhaustion can produce SKIPPED run status. Sources: openhands:gap-automations.
Retries [unknown] Retries: The November first-party documentation and repository pass did not establish unattended triggers at the rubric's required scope. No installed-product fixture was run.
Client independence [partial] Automations keep running when the laptop is shut only if the backend runs on a server or cloud host; a laptop-local backend stops with the laptop. Sources: openhands:b09-readme.
Canvas supports goal interruption and preserves failed-message retry controls.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
10 atomic answers
Pause [unknown] Pause: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Stop [unknown] Stop: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Resume [unknown] Resume: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Steer [supported] Normal message interrupts an active /goal and returns control to the user. Sources: openhands:gap-conversations.
Notifications [unknown] Notifications: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Pending decisions [unknown] Pending decisions: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Escalation [unknown] Escalation: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Visible errors [supported] Failed message and Retry control persist after history reload. Sources: openhands:gap-conversations.
Chat terminal parity [unknown] Chat terminal parity: The November first-party documentation and repository pass did not establish intervention and attention at the rubric's required scope. No installed-product fixture was run.
Retained history [supported] Conversation retains message/tool history and supports archival and transcript export. Sources: openhands:gap-conversations.
Canvas shows linked children and grouped conversation metadata. Fleet fixture remains untested.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
8 atomic answers
Fleet fixture [unknown] Fleet fixture: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
Parent child [supported] Child conversations remain linked to their parent. Sources: openhands:gap-conversations.
Dependencies [unknown] Dependencies: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
Attention [unknown] Attention: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
History [supported] Conversation tags store workspace/repo/branch metadata server-side for other browsers. Sources: openhands:gap-conversations.
Workload [unknown] Workload: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
Navigation errors [unknown] Navigation errors: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
Fixture mode [unknown] Fixture mode: The November first-party documentation and repository pass did not establish fleet and epic management ui at the rubric's required scope. No installed-product fixture was run.
Canvas has reusable profiles, goals and backend-specific configuration.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
10 atomic answers
Quick task [supported] New conversation starts a single agent session on the active backend. Sources: openhands:gap-conversations.
Growing epic [unknown] Growing epic: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Presets [unknown] Presets: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Natural language [unknown] Natural language: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Agent configuration [unknown] Agent configuration: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
In run changes [supported] User message interrupts a goal; conversation branching preserves earlier context on supporting local backends. Sources: openhands:gap-conversations.
Live roles [unknown] Live roles: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Retained relationships [unknown] Retained relationships: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Discoverability [unknown] Discoverability: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Auth friction [unknown] Auth friction: The November first-party documentation and repository pass did not establish workflow flexibility and orchestration visibility at the rubric's required scope. No installed-product fixture was run.
Self hosted [unknown] Self hosted: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Provisioning [unknown] Provisioning: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Payer [unknown] Payer: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Resources [unknown] Resources: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Region [unknown] Region: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Expiry [unknown] Expiry: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Client disconnect [partial] Vendor states remote server agents continue with laptop off; disconnect was not tested. Sources: openhands:nov07-canvas.
Other device continuation [unknown] Other device continuation: The November first-party documentation and repository pass did not establish laptop-independent execution at the rubric's required scope. No installed-product fixture was run.
Optional containers and worktrees separate execution/files, with documented shared-HOME and shared-workspace limits.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
12 atomic answers
Checkout [unknown] Checkout: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Worktree [supported] Local child conversations may use an isolated worktree or parent shared workspace. Sources: openhands:gap-conversations.
Container [supported] Optional container per new conversation. Sources: openhands:nov07-readme.
Vm [unknown] Vm: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Os sandbox [unknown] Os sandbox: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Filesystem [partial] Direct mode has host filesystem access; same host workspace mounts still share files. Sources: openhands:nov07-readme.
Process [unknown] Process: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Network [unknown] Network: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Secrets [unknown] Secrets: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Code separation [unknown] Code separation: The November first-party documentation and repository pass did not establish workspace and process isolation at the rubric's required scope. No installed-product fixture was run.
Security boundary [partial] Docker execution boundary does not isolate the whole Canvas/automation stack. Sources: openhands:nov07-readme.
Canvas reloads backend conversation history and keeps archived conversations without deleting them.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
12 atomic answers
Client disconnect [unknown] Client disconnect: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
Worker crash [partial] The Docker container exits with code 1 after the Agent Server fails /alive for about 30 seconds; it recovers only with a configured restart policy. Sources: openhands:b09-readme.
Service restart [unknown] Service restart: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
Expiry [unknown] Expiry: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
Transcript reload [supported] Full archived conversation history remains on backend; missing runtimes remain readable but cannot be restored from conversation. Sources: openhands:gap-conversations.
Process reconnect [unknown] Process reconnect: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
Task resume [unknown] Task resume: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
State reconstruction [partial] Mounted workspace and conversation state survive container replacement. Sources: openhands:nov07-readme.
Context loss [supported] Runtime-missing conversation is shown archived; readable history does not imply a recoverable execution process. Sources: openhands:gap-conversations.
Duplicate actions [unknown] Duplicate actions: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
Manual repair [partial] Container health failure exits visibly; restart requires a configured restart policy. Sources: openhands:nov07-readme.
Multi day continuity [unknown] Multi day continuity: The November first-party documentation and repository pass did not establish recovery, expiry and continuity at the rubric's required scope. No installed-product fixture was run.
The user chooses workspace mounts and runtime configuration.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
10 atomic answers
Dependencies [supported] Local Node/uv or Docker prerequisites. Sources: openhands:nov07-readme.
Caches [unknown] Caches: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Snapshots [unknown] Snapshots: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Services [unknown] Services: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Ports [unknown] Ports: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Secrets [unknown] Secrets: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Duplicate installs [unknown] Duplicate installs: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Collisions [unknown] Collisions: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Remote setup [supported] Operator provisions and hardens the server. Sources: openhands:nov07-readme.
Automatic repair [unknown] Automatic repair: The November first-party documentation and repository pass did not establish development environment reliability at the rubric's required scope. No installed-product fixture was run.
Goals and experimental critic refinement have explicit iteration limits.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
8 atomic answers
Time limits [unknown] Time limits: The November first-party documentation and repository pass did not establish limits and failure handling at the rubric's required scope. No installed-product fixture was run.
Turn limits [supported] Experimental OpenHands critic refinement defaults to at most three attempts; not generalized to ACP agents. Sources: openhands:gap-critic.
Token limits [unknown] Token limits: The November first-party documentation and repository pass did not establish limits and failure handling at the rubric's required scope. No installed-product fixture was run.
Spend limits [supported] Enterprise managed gateway supports monthly organization/user budgets. Direct provider calls are outside enforcement. Sources: openhands:gap-budgets.
Timeouts [unknown] Timeouts: The November first-party documentation and repository pass did not establish limits and failure handling at the rubric's required scope. No installed-product fixture was run.
Retries [unknown] Retries: The November first-party documentation and repository pass did not establish limits and failure handling at the rubric's required scope. No installed-product fixture was run.
Provider failures [unknown] Provider failures: The November first-party documentation and repository pass did not establish limits and failure handling at the rubric's required scope. No installed-product fixture was run.
Enforcement [supported] Refinement stops at score threshold or max iterations; documentation only. Sources: openhands:gap-critic.
Canvas displays automation phase/status and exports reported costs with null for unavailable cost.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
13 atomic answers
Logs [supported] Script-run View logs shows the script's output; on cloud backends logs are read from the run sandbox and become unavailable after that sandbox is deleted. Sources: openhands:gap-automations.
Traces [unknown] Traces: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Failure reasons [partial] Health check can exit the container after roughly thirty seconds of Agent Server failure. Sources: openhands:nov07-readme.
Usage export [supported] Activity Log exports CSV/JSON with numeric reported cost or null. Sources: openhands:gap-automations.
Task stats [unknown] Task stats: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Agent stats [unknown] Agent stats: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Project stats [unknown] Project stats: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Time stats [unknown] Time stats: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Failure status [supported] Statuses include failed/cancelled/skipped and structured disablement reason with time. Sources: openhands:gap-automations.
Approval status [unknown] Approval status: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Reconnect consistency [unknown] Reconnect consistency: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Performance method [unknown] Performance method: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Responsiveness [unknown] Responsiveness: The November first-party documentation and repository pass did not establish truthful status and operational statistics at the rubric's required scope. No installed-product fixture was run.
Canvas automation history exposes reported per-run LLM cost with explicit missing values.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
14 atomic answers
Task cost [supported] Completed automation run reported LLM cost displayed in USD to four decimal places. Sources: openhands:gap-automations.
Session cost [unknown] Session cost: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Child cost [unknown] Child cost: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Epic cost [unknown] Epic cost: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Harness aggregation [unknown] Harness aggregation: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Billed vs estimated [partial] Value is described as backend-reported LLM cost; the Activity Log does not say whether it is billed or estimated, and no provider invoice reconciliation is established. Sources: openhands:gap-automations.
Tokens [unknown] Tokens: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Subscription allowance [unknown] Subscription allowance: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Provider reconciliation [partial] Enterprise gateway budget counters and conversation-ledger estimates have different scope/period; docs explain reconciling those two views, not reconciliation with provider invoices. Sources: openhands:gap-budgets.
Compute cost [unknown] Compute cost: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Refresh delay [supported] Enterprise budget spend is marked live, stale or unavailable, with observation time; unavailable never means zero. Sources: openhands:gap-budgets.
Alerts [supported] Enterprise organization thresholds such as 80/90/100 percent notify Admins/Owners by email and configured Slack. Sources: openhands:gap-budgets.
Export [supported] CSV/JSON activity export includes numeric cost or null. Sources: openhands:gap-automations.
Drilldown [unknown] Drilldown: The November first-party documentation and repository pass did not establish per-task api and compute cost at the rubric's required scope. No installed-product fixture was run.
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
12 atomic answers
Concurrency [supported] Automation run can be SKIPPED when backend concurrency limit is reached. Sources: openhands:gap-automations.
Queues [unknown] Queues: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Cpu ram [unknown] Cpu ram: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Backpressure [supported] Capacity limitation is surfaced as skipped status rather than assumed successful execution. Sources: openhands:gap-automations.
Host awareness [unknown] Host awareness: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Check reuse [unknown] Check reuse: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Check deduplication [unknown] Check deduplication: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Bounded tests [unknown] Bounded tests: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Cancellation [unknown] Cancellation: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Fairness [unknown] Fairness: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Recovery [unknown] Recovery: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Shared limits [unknown] Shared limits: The November first-party documentation and repository pass did not establish compute-aware scheduling and verification at the rubric's required scope. No installed-product fixture was run.
Cloud organization automation conversations support read-only team visibility. Same-agent co-prompting remains unestablished.
Scope
Surfaces: OpenHands Cloud · Version: Not specified · Plan: Cloud organization automation conversations · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
11 atomic answers
Visibility [supported] Other organization users may follow automation conversation history read-only. Sources: openhands:gap-automations.
Discussion [unknown] Discussion: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Comments [unknown] Comments: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Presence [unknown] Presence: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Coprompting [unsupported] Non-owner viewers of shared automation conversations explicitly cannot send messages. Scope is shared automation runs, not every collaboration mode. Sources: openhands:gap-automations.
Ordering [unknown] Ordering: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Conflicts [unknown] Conflicts: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Attribution [unknown] Attribution: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Permissions [unknown] Permissions: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Plan restrictions [unknown] Plan restrictions: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Owner online [unknown] Owner online: The November first-party documentation and repository pass did not establish live human collaboration at the rubric's required scope. No installed-product fixture was run.
Surfaces: OpenHands Cloud, OpenHands Agent Server backends · Version: Not specified · Plan: Cloud organization roles; Teams/Enterprise self-hosted SSO/SAML · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
9 atomic answers
Organization roles [supported] Member creates conversations/private settings; Admin invites/promotes and manages org settings; Owner deletes org/claims Git organizations. Sources: openhands:gap-roles.
Project roles [unknown] Project roles: The November first-party documentation and repository pass did not establish roles and access boundaries at the rubric's required scope. No installed-product fixture was run.
Task roles [unknown] Task roles: The November first-party documentation and repository pass did not establish roles and access boundaries at the rubric's required scope. No installed-product fixture was run.
Membership [supported] Members join as Member; Admins invite and change Member/Admin roles. Sources: openhands:gap-roles.
Sso [supported] OpenHands Enterprise installations document SAML single sign-on with a corporate identity provider; the pricing page lists SSO/SAML on Teams and Enterprise, not free Agent Canvas. Sources: openhands:src1-sso, openhands:src1-pricing.
Provisioning [unknown] Provisioning: The November first-party documentation and repository pass did not establish roles and access boundaries at the rubric's required scope. No installed-product fixture was run.
Unauthorized access [unknown] Unauthorized access: The November first-party documentation and repository pass did not establish roles and access boundaries at the rubric's required scope. No installed-product fixture was run.
Revocation [unknown] Revocation: The November first-party documentation and repository pass did not establish roles and access boundaries at the rubric's required scope. No installed-product fixture was run.
Plan scope [supported] Cloud/Enterprise organization roles; free Canvas uses a shared identity. Sources: openhands:gap-roles.
Self-hosted public mode requires the backend API key.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
14 atomic answers
Command [unknown] Command: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Tool [unknown] Tool: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Network [unknown] Network: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Secret [partial] X-Session-API-Key authenticates API calls; this is server access, not per-tool approval. Sources: openhands:nov07-selfhost.
Delivery [unknown] Delivery: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Approver identity [unknown] Approver identity: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Lifetime [unknown] Lifetime: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Allow [unknown] Allow: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Deny [unknown] Deny: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Revoke [unknown] Revoke: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Tunnels [unknown] Tunnels: The November first-party documentation and repository pass did not establish approvals and temporary access at the rubric's required scope. No installed-product fixture was run.
Defaults [supported] Public mode does not inject the key into frontend HTML. Sources: openhands:nov07-selfhost.
Canvas exports timestamped conversation/tool history and retains archived messages.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
11 atomic answers
Actions [unknown] Actions: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Approvals [unknown] Approvals: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Edits [unknown] Edits: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Discussion [unknown] Discussion: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Decisions [unknown] Decisions: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Cross user persistence [unknown] Cross user persistence: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Cross session persistence [supported] Archiving preserves full backend history and is reversible. Sources: openhands:gap-conversations.
Export [supported] Self-contained Markdown/HTML export optionally includes full tool input/output and timestamps; generated in browser. Sources: openhands:gap-conversations.
Mutability [unknown] Mutability: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Retention [supported] Archive preserves history until deletion; Delete all includes hidden archived conversations. Sources: openhands:gap-conversations.
Integrity [unknown] Integrity: The November first-party documentation and repository pass did not establish decision and activity history at the rubric's required scope. No installed-product fixture was run.
Canvas exposes combined commit/uncommitted diffs and experimental critic feedback.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
13 atomic answers
Pending reviews [unknown] Pending reviews: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Change understanding [supported] Unified Commits drawer combines commit history and uncommitted changes. Sources: openhands:gap-conversations.
Comments [unknown] Comments: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Request changes [unknown] Request changes: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Shipping guidance [unknown] Shipping guidance: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Effort [unknown] Effort: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Missed evidence [unknown] Missed evidence: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Keyboard [unknown] Keyboard: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Narrow screen [unknown] Narrow screen: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Context switching [unknown] Context switching: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Roles [unknown] Roles: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Retained context [unknown] Retained context: The November first-party documentation and repository pass did not establish review experience and guidance at the rubric's required scope. No installed-product fixture was run.
Canvas renders file artifacts, images and unified code-change views.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends, OpenHands Cloud · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
13 atomic answers
Diffs [supported] Commits drawer combines commit history and uncommitted changes. Sources: openhands:gap-conversations.
Documents [supported] Created Markdown files render inline with full-file Files drawer access. Sources: openhands:gap-conversations.
Screenshots [unknown] Screenshots: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Recordings [unknown] Recordings: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Previews [partial] Browser view is rendered; previewing a running app in it was not established. Sources: openhands:nov07-architecture.
Test outputs [unknown] Test outputs: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Reviewer outputs [unknown] Reviewer outputs: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Gallery [unknown] Gallery: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Task links [unknown] Task links: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Revision links [unknown] Revision links: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Persistence [unknown] Persistence: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Child findings [supported] Child conversation results return to the parent conversation. Sources: openhands:gap-conversations.
Prompt leakage [unknown] Prompt leakage: The November first-party documentation and repository pass did not establish retained artifacts and outputs at the rubric's required scope. No installed-product fixture was run.
Experimental critic evaluates OpenHands-agent results and can request bounded refinement.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
11 atomic answers
Tests [partial] The Issue-to-PR automation prompt has the agent run tests before opening the PR; no enforced test gate is documented. Sources: openhands:b09-issue-pr.
Lint [unknown] Lint: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Types [unknown] Types: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Ci [unknown] Ci: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Independent review [supported] Additional critic evaluation supplies likelihood score and issue categories; no human review or acceptance-test equivalence. Sources: openhands:gap-critic.
Spec checks [unknown] Spec checks: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Required checks [unknown] Required checks: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Failure blocking [unknown] Failure blocking: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Requirement conflicts [unknown] Requirement conflicts: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Configured policy [supported] User can enable refinement and configure threshold/max attempts in Verification settings. Sources: openhands:gap-critic.
Enforced policy [unknown] Enforced policy: The November first-party documentation and repository pass did not establish verification and spec gates at the rubric's required scope. No installed-product fixture was run.
Canvas Git controls prompt the agent to commit, pull, push and work with PRs.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
9 atomic answers
Branches [supported] Issue-to-PR creates a feature branch per issue with a configurable prefix; Cloud child conversations use the selected branch. Sources: openhands:b09-issue-pr, openhands:b09-conversations.
Worktrees [supported] Local child conversations can use an isolated worktree or the parent's shared workspace. Sources: openhands:b09-conversations.
Pull requests [supported] Git-actions control sends pull-request prompts to the agent; no completed PR fixture. Sources: openhands:gap-conversations.
Comments [unknown] Comments: The November first-party documentation and repository pass did not establish git and pull request workflow at the rubric's required scope. No installed-product fixture was run.
Ci response [unknown] Ci response: The November first-party documentation and repository pass did not establish git and pull request workflow at the rubric's required scope. No installed-product fixture was run.
Rebase [unknown] Rebase: The November first-party documentation and repository pass did not establish git and pull request workflow at the rubric's required scope. No installed-product fixture was run.
Conflicts [unknown] Conflicts: The November first-party documentation and repository pass did not establish git and pull request workflow at the rubric's required scope. No installed-product fixture was run.
Product state [unknown] Product state: The November first-party documentation and repository pass did not establish git and pull request workflow at the rubric's required scope. No installed-product fixture was run.
Surfaces: Agent Canvas · Version: First-party sources checked 2026-11-07; pinned implementations retain their cited revision. · Plan: Not specified · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-07
The remaining atomic questions and the rubric scenario are unverified. Documentation is not a successful product test.
6 atomic answers
Client license [supported] MIT, subject to component notices. Sources: openhands:nov07-license.
Server license [unknown] Server license: The November first-party documentation and repository pass did not establish source and license at the rubric's required scope. No installed-product fixture was run.
Hosted extensions [unknown] Hosted extensions: The November first-party documentation and repository pass did not establish source and license at the rubric's required scope. No installed-product fixture was run.
Hosted parity [unknown] Hosted parity: The November first-party documentation and repository pass did not establish source and license at the rubric's required scope. No installed-product fixture was run.
Export [supported] Conversation transcript downloads as Markdown/HTML. Sources: openhands:gap-conversations.
Import [unknown] Import: The November first-party documentation and repository pass did not establish hosting and exit at the rubric's required scope. No installed-product fixture was run.
Cancellation [unknown] Cancellation: The November first-party documentation and repository pass did not establish hosting and exit at the rubric's required scope. No installed-product fixture was run.
Account independence [unknown] Account independence: The November first-party documentation and repository pass did not establish hosting and exit at the rubric's required scope. No installed-product fixture was run.
History portability [supported] Local export loads full event history; Cloud export uses currently loaded events. Sources: openhands:gap-conversations.
Backend settings/secrets and browser-generated exports have explicit boundaries. At v1.38.0 an anonymous usage-data switch defaults on with a consent prompt and can be turned off on local and self-hosted backends; it is forced on for OpenHands Cloud.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: Telemetry from OpenHands v1.38.0 source (e14b76c); other parts from first-party documentation checked 2026-11-07. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
15 atomic answers
Code data [unknown] Code data: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Prompts [unknown] Prompts: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Credentials [supported] Agent Server owns ACP credentials/subprocesses; onboarding secrets go to backend secret store. Sources: openhands:gap-acp.
Logs [unknown] Logs: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Artifacts [unknown] Artifacts: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Retention [unknown] Retention: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Deletion [unknown] Deletion: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Training [unknown] Training: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Residency [unknown] Residency: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Subprocessors [unknown] Subprocessors: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Telemetry disclosure [unknown] Telemetry disclosure: The November first-party documentation and repository pass did not establish data handling and telemetry at the rubric's required scope. No installed-product fixture was run.
Telemetry defaults [supported] The anonymous usage-data switch defaults on when the user has not chosen; on a Cloud backend it is forced on. Source inspection at v1.38.0, not tested. Sources: openhands:src1-telemetry-settings.
Telemetry controls [supported] App settings offer a switch to stop sending anonymous usage data on local and self-hosted backends; it is disabled on Cloud backends. Source inspection at v1.38.0, not tested. Sources: openhands:src1-telemetry-settings.
Consent prompt [supported] A consent modal asks whether to send anonymous usage data and records the choice. Source inspection at v1.38.0; when it appears was not tested. Sources: openhands:src1-telemetry-consent.
Provider separation [supported] ACP provider owns its LLM/tools/execution; Canvas sends turns and renders results. Sources: openhands:gap-acp.
The current public self-hosted offer is free Canvas, Teams at $500/month for ten users plus $50/month additional seats with a 5000-session monthly cap, and custom Enterprise. No one-time licence is listed. The optional Cloud backend is not priced in this table; its documented OpenHands LLM credits buy model access, so product-wide metered ADE and compute pricing remain unestablished.
Scope
Surfaces: Agent Canvas, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
generally-available
Delivery
native
Checked
2026-11-07 (15 days old)
Agent cross-check
Not separately cross-checked
Only the named atomic answers are established by this source pass. Remaining atomic answers retain their documented gaps.
Interval [supported] Teams billed monthly at $500 for up to ten users; $50 monthly additional seats. Sources: openhands:gap-pricing.
Seats [supported] Teams includes up to ten users; Enterprise unlimited seats. Sources: openhands:gap-pricing.
Seat minimum [unknown] Seat minimum: The November first-party documentation and repository pass did not establish price and included usage at the rubric's required scope. No installed-product fixture was run.
Workspace fees [unknown] Workspace fees: The November first-party documentation and repository pass did not establish price and included usage at the rubric's required scope. No installed-product fixture was run.
Included usage [supported] The current Teams table caps usage at 5000 sessions per month; Enterprise is uncapped. Sources: openhands:gap39-plan-table.
Credits [unknown] Credits: The November first-party documentation and repository pass did not establish price and included usage at the rubric's required scope. No installed-product fixture was run.
Overages [unknown] Overages: The November first-party documentation and repository pass did not establish price and included usage at the rubric's required scope. No installed-product fixture was run.
Taxes [unknown] Taxes: The November first-party documentation and repository pass did not establish price and included usage at the rubric's required scope. No installed-product fixture was run.
Model cost [supported] Self-hosted plans use user-provided LLM keys. Optional OpenHands Cloud LLM keys consume separately purchased model credits and can also be used in other agents. Sources: openhands:gap39-plan-table, openhands:gap39-cloud-model-credits.
Teams and Enterprise plans carry paid identity governance; support ranges from community support for Agent Canvas to best-effort email on Teams and SLA-based response on Enterprise.
Scope
Surfaces: Agent Canvas, OpenHands Cloud, OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. Batch 09 recheck 2026-11-22: docs.openhands.dev Agent Canvas pages and repository at v1.43.0 (d5d7908, released 2026-11-20); release list v1.37.0-v1.43.0. Newer code is not attributed to earlier releases. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-22 (checked today)
Agent cross-check
2026-11-22
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
Documentation and public-code recheck only (2026-11-22); no product fixture was run.
Support [supported] Canvas community support; Teams best-effort email; Enterprise SLA-based response times. Sources: openhands:gap-pricing.
Response commitments [unknown] Response commitments: The November first-party documentation and repository pass did not establish availability and support at the rubric's required scope. No installed-product fixture was run.
Release activity [supported] Seven releases from v1.37.0 (2026-11-06) to v1.43.0 (2026-11-20). Sources: openhands:b09-releases.
Status history [unknown] Status history: The November first-party documentation and repository pass did not establish availability and support at the rubric's required scope. No installed-product fixture was run.
Deprecation [unknown] Deprecation: The November first-party documentation and repository pass did not establish availability and support at the rubric's required scope. No installed-product fixture was run.
Export options [unknown] Export options: The November first-party documentation and repository pass did not establish availability and support at the rubric's required scope. No installed-product fixture was run.
Trial [unknown] Trial: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Credits trial [unknown] Credits trial: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Self host cost [partial] Self-hosted VM is labeled free, OSS and self-managed. The page gives no infrastructure or provider cost detail. Sources: openhands:nov07-canvas.
Paid only [unknown] Paid only: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Early access [unknown] Early access: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Free scenarios [unknown] Free scenarios: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Card required [unknown] Card required: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Usage caps [unknown] Usage caps: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Paid evaluation [unknown] Paid evaluation: The November first-party documentation and repository pass did not establish useful free tier and evaluation access at the rubric's required scope. No installed-product fixture was run.
Pricing page publishes monthly Teams rate and session caps, with Enterprise by quote.
Scope
Surfaces: OpenHands Agent Server backends · Version: First-party documentation/source checked 2026-11-07; exact pinned source revision is recorded in each citation. No installed-product fixture. · Plan: Agent Canvas unless answer explicitly scopes paid Teams/Enterprise · OS: Not specified · Execution modes: Not specified · Region: Not specified · Harness: Not specified
Availability
unknown
Delivery
unknown
Checked
2026-11-07 (15 days old)
Agent cross-check
2026-11-10
Only the named atomic facts are established by documentation. Remaining rubric scenarios and unknown answers were not demonstrated.
9 atomic answers
Units [supported] Teams $500 monthly up to ten users plus $50 monthly extra seats. Sources: openhands:gap-pricing.
Included limits [supported] Teams 5000 sessions per month; Enterprise uncapped. Sources: openhands:gap-pricing.
Overages [unknown] Overages: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Compute markup [unknown] Compute markup: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Model markup [unknown] Model markup: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Minimum commitment [unknown] Minimum commitment: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Cancellation [unknown] Cancellation: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Sales only unknowns [supported] Enterprise exact price requires a quote. Sources: openhands:gap-pricing.
Transparency classification [unknown] Transparency classification: The November first-party documentation and repository pass did not establish pricing transparency at the rubric's required scope. No installed-product fixture was run.
Canvas launches local Agent Server/automation services with Node 24 and uv, or optional Docker runtimes. Backends can live on laptop, remote VM or managed cloud. Same host workspace mounts still share files.
Kind
documented
Locator
Overview, setup, features and component boundaries
The public native Expo MVP can list conversations, stream history/events and send messages to a reachable Agent Server using a session API key. Source-build instructions cover Android and iOS with native toolchains. Cloud login, QR pairing and a laptop remote-session facade are still planned; no store/release binary is asserted.
The native mobile client implements REST conversation listing/history and sends message events to /api/conversations/{id}/events; ChatScreen calls that sendMessage path, and useConversationSocket streams events through WebSocket. app.json declares iOS and Android native platforms.
The product-owned agent-canvas CLI starts the local Agent Server, automation backend and pre-built frontend. It exposes version, stack-info, public/frontend-only/backend-only and help flags. This establishes a user-runnable launcher CLI, without asserting a separate task-operation CLI.
Local ACP agents reuse provider-native logins; clean containers/cloud need credentials. Pi/OpenCode are offered only on local/self-hosted backends. Same-provider conversations sharing container HOME can race on auth/config/lock files; Canvas cannot yet send SDK per-conversation data-dir isolation. Agent choice is per backend and running conversations retain their original agent.
Kind
documented
Locator
Authentication; Per-conversation isolation; Model switching
Agent Canvas runs the OpenHands agent and third-party agents such as Claude Code, Codex, Gemini CLI, Pi and OpenCode, or any ACP agent. Agent Canvas lists 'Bring your own model: use with any LLM' and automations that integrate with Slack, GitHub and Linear and run on a schedule or in response to webhook events. Running on a cloud server lets agents continue when the laptop is shut. The Docker entrypoint stops the container with exit code 1 when the Agent Server fails its /alive check for about 30 seconds; automatic recovery needs a restart policy such as --restart unless-stopped.
Bring your own LLM and API key: any model supported by litellm; basic settings list popular providers (OpenHands provider, Anthropic, OpenAI and others); advanced settings take a provider-prefixed custom model and base URL. LLM profiles save multiple configurations and switch between them during an active conversation.
Kind
documented
Locator
Basic settings; Advanced settings (Custom Model, Base URL); LLM Profiles
The npm package's only bin, agent-canvas, starts the full local stack or a frontend-only or backend-only split, with port, bind host, public API-key mode and version/info flags. Its help says LLM settings are configured in the web UI. It has no commands to start, list or steer conversations.
Kind
source-inspected
Locator
bin/agent-canvas.mjs lines 55-100 (--help text); package.json bin entry agent-canvas
Automations trigger on a schedule, webhook or event. Event triggers accept GitHub, Jira Data Center, Bitbucket Data Center and Azure DevOps events; custom webhook sources registered under Automate > Events get a webhook URL and signing secret, when the backend's automation service supports it.
Kind
documented
Locator
Trigger configuration; Event triggers and event types; Event Sources
Canvas loads workspace/user persistent memory when enabled. Skills/MCP/secrets are backend-scoped. MCP health distinguishes reachable, rejected credentials and connection failure; public-server reachability does not verify credentials. Skills load on new conversation start.
Prebuilt automation that watches a Slack channel and triggers an OpenHands agent when a message matches the instructions; requires creating a Slack app and installing MCP servers and secrets in Agent Canvas.
Issue-to-PR automations watch GitHub Issues, GitLab Issues, Jira or Linear for issues labeled or moved to an implementation-ready state (polled, default every 15 minutes), read the issue title, description, acceptance criteria and linked dependencies, create a feature branch (configurable prefix), run tests, open a PR linking the issue and post progress and the PR link back to the tracker. The steps are carried out by the agent following the automation prompt.
Kind
documented
Locator
How it works steps 1-6; supported combinations; setup (trigger label or state, branch prefix, schedule)
The Canvas Task List derives its entries from the latest TaskTrackerObservation in the current conversation event store. IDs are generated from list position and statuses are todo/in_progress/done. Inspection of the native task-list route and API/store surfaces found no independent issue/backlog work-item entity; automations are separate trigger definitions.
Canvas child conversations stay linked and return results; local children may use isolated worktrees or shared workspace. Commit drawer shows diffs and agent Git/PR actions. Goals use a judge with round caps and user-message interruption. Transcripts export locally as Markdown/HTML with tool details; archive keeps backend history.
launch_child_conversation starts a child on a local or Cloud target; local children use an isolated worktree or the parent's shared workspace; Cloud children use the selected repository and branch. The child stays linked to its parent and its result returns to the parent conversation. While an agent works, its current stage appears as a live group at the end of the chat; this is the only live agent status shown.
Kind
documented
Locator
Child Conversations; stage groups; Live Agent Status
Automations have pending/running/completed/failed/cancelled/skipped status; concurrency limits may skip runs. Activity exports CSV/JSON with numeric USD cost or null. Automations export/import versioned JSON; disabling preserves configuration. Other Cloud organization users can view run conversations read-only but cannot send messages.
Kind
documented
Locator
Run statuses; Activity costs; Shared automation conversations; Export/import
Experimental critic applies to OpenHands-agent conversations; ACP agents may not expose the path. Default threshold 0.6 and maximum three refinement attempts. Results expose score and detected issue categories; this is additional evaluation, not a successful acceptance-test result.
Enterprise managed-gateway budgets cap organization/user monthly spend. Gateway admits requests and can block at limits; direct provider calls are outside enforcement. Dashboard conversation estimates and gateway budget counters differ in scope/freshness. Spend may be live/stale/unavailable, not zero when unavailable; threshold alerts go to admins/owners.
Cloud/Enterprise organization roles are Member, Admin and Owner. Members create conversations and own private MCP/secrets/settings; Admins invite/promote users and manage LLM settings/credits; Owners delete organizations and claim Git organizations. Private settings remain private from other members including admins.
Canvas is free, MIT, single-user/shared-identity and self-hosted. Teams is $500 monthly for up to ten users with $50 monthly per additional seat, 5000 sessions per month, 30-day trial, RBAC and SSO/SAML. Enterprise has custom pricing. LLM keys are user-provided.
The guide explains how to let users of an OpenHands Enterprise installation sign in through a corporate SAML identity provider such as Okta, Entra ID, Google Workspace or ADFS.
Agent Canvas is free and MIT licensed. Teams costs 500 USD per month for up to ten users plus 50 USD per extra seat and lists SSO/SAML and RBAC; Enterprise has custom pricing. The page returns HTTP 403 to some automated fetchers but loads with a browser user agent.
Kind
documented
Locator
Plan cards and comparison table (Agent Canvas, Teams, Enterprise)
App settings render a 'send anonymous usage data' switch that defaults on and saves user_consents_to_analytics and the telemetry consent as granted or denied. On a Cloud backend the switch is disabled and stays on.
Kind
source-inspected
Locator
src/routes/app-settings.tsx lines 112-136 and 217-231 (enable-analytics-switch), release v1.38.0 commit e14b76c
The current public price table lists free self-hosted Canvas, self-hosted Teams at $500/month for ten users plus $50/month additional seats, capped at 5000 sessions monthly, and custom Enterprise. It lists no one-time licence. The separately linked Cloud backend is not priced in this table.
Kind
documented
Locator
Agent Canvas, Teams and Enterprise; Comparison table
Cloud users must buy at least $10 in credits before generating an OpenHands LLM Key. That key can be used in the CLI, self-hosted OpenHands or other coding agents. These are model-provider credits, not proof of a metered ADE or sandbox price.