A source-linked comparison across the complete Podium ADE rubric. 48 criteria sourced for both products; 37 criteria also agent cross-checked for both. 51 rows have at least one sourced finding, including 51 with an agent-cross-checked finding. No score or winner is inferred.
Each row shows a finding's verification status, support, check date and answer. Its scope, limitations, atomic answers and sources are on the product's page. 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.
Amp offers Mac (macOS 26+, Apple silicon), Windows (10 1809+ or 11, x64 and Arm) and native Linux desktop apps (Debian 12+, Ubuntu 22.04+, Fedora, Arch; x86-64 or ARM64); Linux includes a local systemd runner. iPhone and iPad (iOS 26+) use a public TestFlight beta and Android 13+ a direct APK; no store release is documented. Web threads run in Orbs at ampcode.com, and the CLI runs locally or in scripts. Windows notifications report ready threads, mentions and access requests. No dedicated editor extension: the CLI connects to VS Code, Cursor, Windsurf, Zed and Neovim through ide connect (Neovim needs the Amp plugin). The JetBrains integration is deprecated and gets no updates.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Podium ships desktop, CLI and browser clients, including phone-browser control. The native iPhone app is private beta; the complete download offering contains no native Android app.
Amp supplies its own coding-agent harness across clients and builds prompts and tools around the selected model; custom agent modes and subagents remain Amp agents. With a linked Claude subscription, Amp also offers a Claude Code agent under Your Agents that runs Claude Code with Amp's tools, and Claude models served by that plan run through Claude Code via the Claude Agent SDK on Amp's servers. The agent is fixed per thread once started on Claude Code or a plugin agent. Other third-party coding CLIs are not documented as selectable harnesses in the checked pages.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium launches the native Claude Code, Codex, Grok, OpenCode, Cursor and Pi CLIs in persistent terminals on machines you control. The 0.2.0 release calls Pi fully supported. Models come from each harness catalog, and capabilities differ per harness: Pi declares no quota, handoff or hooks. Mixed-harness operation was not exercised.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
The pricing guide supports linking ChatGPT and Claude subscriptions. BYOK is a separate inference route. Further scoped facts are recorded in the atomic answers. Fable BYOK needs an Anthropic workspace with data retention; Sign in with ChatGPT connects each ChatGPT account to only one Amp account, while ChatGPT Subscription (Legacy) lets one ChatGPT account serve several Amp accounts; Copilot needs Copilot CLI enabled and *.ghe.com hosts are unsupported; Gemini image generation needs an AI Studio key; one active connection per subscription type. An Amp account is required; Hobby has no monthly fee.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium supports native coding-agent logins and managed provider credentials. Claude subscriptions (via native login or a managed setup-token) and local ChatGPT/Codex login are supported paths, alongside Anthropic and OpenAI API keys. Docs list billing and hosted metering as not implemented product claims.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
The quickstart permits a project without a Git repository. Orbs support GitHub and other HTTPS Git hosts; GitHub private access uses an installed GitHub App. Runners start with amp --no-tui on a user machine; self-hosted orbs use amp orb provider with a user dispatch program or the Helm chart for GKE, EKS and AKS. Quickstart: sign up, create a project from a Git repository or start without one, then run an agent in an orb; apps and CLI are optional. Amp prepares project orbs with setup scripts and saves a snapshot per orb size, which later threads restore. Docs cover pre-clone and pre-setup scripts, .agents/setup, .agents/resume and a Setup Log for failed repository setup.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Docs describe a first-run flow (install, podium, register a Git repository, confirm a detected logged-in harness) and a headless VPS installer with machine joining. Setup time and failure rate were not measured, and no self-serve managed cloud setup is documented. The current README and installation guide use podium.do/download and podium.do/install.sh as the canonical stable entrypoints.
A public External API v2 covers workspace thread data, members, projects and repository search, groups, entitlements, model-provider keys and usage analytics through machine-to-machine OAuth apps; it can delete but not create threads. Non-interactive use sets AMP_API_KEY to an sgamp_ access token; service-account keys run as the workspace. amp -x runs one prompt, -ox runs it in an orb, --executor runner:<id> uses a runner; thread, space and sync subcommands exist. Puck is exposed as a remote MCP server at https://ampcode.com/mcp for Claude, ChatGPT and other MCP clients. CLI and SDK start and continue threads, spawn orbs and runner threads, create Spaces, and list, search, label, archive, delete and export threads. TypeScript and Python SDKs run the agent and continue threads by ID; the TypeScript SDK needs an Amp CLI at least as new as its pinned version, a legacy SDK release serves pre-Amp Neo use, and the docs say Amp keeps no backward compatibility. GitHub events continue or spawn threads; plugin webhooks wake orb threads with at-least-once delivery; Amp-hosted repositories offer a post-receive webhook.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The podium CLI covers issue create, show, update, claim, dependencies, artifacts, session send/status/read/stop and review actions, and the server exposes the same issue commands as a bearer-token MCP tool endpoint. No documented public HTTP API, SDK or webhooks were found.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Local commands and remote MCP servers are supported. Remote servers can use OAuth or configured headers. Further scoped facts are recorded in the atomic answers. Plugins register custom tools with amp.registerTool. Repository, personal and workspace AGENTS.md with glob-scoped @-mentions; SKILL.md skills, including Claude-compatible skill directories. Amp provides GitHub, GitLab, Google, Linear and Slack integrations; each person connects their own account and admins install shared apps; other services connect through MCP. TypeScript or JavaScript plugins add tools, commands, event handlers, skills and UI.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium gives agents its own issue tools over CLI and MCP, and agent profiles import and export Claude Code agent files. No Podium documentation for user-configured MCP clients, plugins or custom tools was found.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-07. Scope and sources
Docs call the Mac, Windows and Linux apps native; Windows installs through App Installer for x64 and Arm. No UI framework is named. Only vendor documentation calling the apps native; no technical documentation or source names the framework, and no binary was inspected. Docs call the iOS and Android apps native; no framework is named.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Desktop releases use a Tauri 2 shell that serves the Podium web UI; mobile source uses Expo and React Native. A native iPhone release is described as private beta.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
GitHub webhooks can continue threads or spawn Orbs from issues and pull requests. Linear or GitHub remain the trackers; Amp works the issue in a thread and replies there. Assigning or mentioning Amp in Linear starts a session; spawners start an orb per new GitHub issue, PR or review request. Linear sessions reply in Linear and orbs update issues through the user's connection; PR subscriptions follow reviews. Other tracker edits stay in the tracker. Prompts start work in a new thread, a Space chat or Puck. Threads have stable T-... URLs. Amp comments on Linear issues (as Amp or the user) and updates issues through the user's Linear connection; on GitHub it pushes, opens PRs and comments. Issue and PR subscriptions feed new comments, reviews and state changes into the thread; Linear sessions show Amp progress. No field-level two-way sync.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium issue records are native and authoritative with stable references, CLI creation and editing, and agent-driven filing from conversation. The in-product assistant has Linear search, create and move tools when a key is configured, but no import or two-way tracker sync was found.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Thread artifacts can hold plans and checklists, persisted with the thread. This does not establish acceptance gates. Humans and agents edit the same artifact. Each artifact save creates a revision. Agents merge against a newer artifact revision when a concurrent save prevents their update. Threads and revisioned artifacts are stored on Amp Server and survive client restarts.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Issue brief, acceptance criteria, state paragraph and todos persist on the durable issue record. Release 0.3.0 removed the separate design and notes fields and appended their text to the brief. On the human issue page the title, description and acceptance are editable and the brief is read-only; agents edit the brief and other fields through issue commands.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-11. Scope and sources
A Space groups threads under an optional goal and project (since Nov 6, 2026); it is not a structured epic record. Threads started by a thread move with it into or out of a Space. Existing threads can be added to or moved between Spaces. A thread removed from a Space keeps its history and environment.
Checked 2026-11-12 (10 days old); agent cross-checked 2026-11-12. Scope and sources
Podium
Agent cross-checkedpartial
Podium supports persistent parent and child issues, phases and typed dependency links, including reparenting an existing issue without replacing its identity. Since 0.3.0 issue types are task, phase and decision: epic and other kinds are retired, and any issue with children acts as a parent. Dependency types are fixed to blocked-by, related, discovered-from, supersedes and duplicates. Cycles are refused and docs list an open blocker as a reason an issue cannot start. No mandatory workflow preset applies in the baseline.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Amp loads AGENTS.md in the current directory, parents and relevant subtrees, plus personal and system guidance. Workspace global guidance applies to top-level agents, with explicit exclusions for subagents, Puck and raw custom agents. Ask Amp to update AGENTS.md from the thread. No memory store is documented; agents find and read earlier threads. Repository and Global AGENTS.md, and thread artifacts. Project, personal and workspace skill repositories. Handoff starts a fresh thread with the context that matters; any thread can be referenced by URL.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
v0.3.0 routes Podium issue instructions and relay environment through terminal and engine session launch. Claude stream sessions use appended instructions and boundary callbacks; Codex server sessions use developer instructions, Grok ACP uses rules, and OpenCode uses the prompt system field. Claude and Codex boundary hooks can re-prime after compaction. Transcript mirroring is on by default; full-text indexing needs the command-palette flag, which defaults off. Agent profiles can carry markdown instructions. Curated memory, skills and memory correction or deletion controls remain unestablished. No instruction-delivery or compaction scenario was run.
Finder returns file paths and line ranges; answers are not documented as citing sources. The Task subagent has a code-search skill; the finder plugin returns files and line ranges. The librarian clones and explains other repositories; Parallel powers web search. The security reference states the client and server do not store, clone or index the entire codebase. System models analyze images, PDFs, audio and video; attachments feed them.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Not established. Podium mirrors daemon transcripts by default, keeps a full-text index only when the command-palette flag is on, and stores session attachments. No repository code search, architecture awareness or API retrieval feature was found.
Threads archive, unarchive and snooze; a separate issue entity is not documented in the checked pages. A Space starts a thread per request; storing ideas separately from threads is not documented in the checked pages. No tracker is required; prompts start work directly. Activity filters threads by time, status, author and project; a separate issue entity to filter is not documented in the checked pages (Threads, Spaces, docs overview). Threads take labels searchable with label:. Activity search and agent thread search cover threads, not issues. A thread is both the work item and the agent conversation.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium exposes native issue creation with an optional start flag, separate dispatch, editing, comments, labels, assignment, search, board filters, archive, sibling ordering and an issue tree. Issue records have their own identity separate from sessions and stages are durable. The idea-to-mission scenario was not run. v0.3.0 refuses an agent close or done update while issue sessions are computing unless --confirm-interrupt is supplied; guidance says to request review and close only on user instruction.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Space cards show each thread’s live status and whether it needs someone. Agents start child threads that report results back; children nest under the parent. Agents and Puck start threads. The Read Thread subagent and @T- mentions read other threads. Delegated threads bring results back into the current thread. Agents and humans use the same T-... thread IDs across clients. Puck archives threads and agents can snooze them on request; other state fields are not documented.
Checked 2026-11-12 (10 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
CLI and MCP address the same canonical issue records for reads, creation, edits, claims, state, dependencies and artifacts. Relayed agents have worker capability scoped to their issue subtree, with --outside-scope confirmation for other existing issues. attachSession is not subtree-gated, but direct rehoming from a real issue is refused and an agent attach naming a Proposed target is refused. That does not make every Proposed subtree inert: operator attachment or spawn-time cwd binding can leave a session attached to a Proposed issue; newSubissue without targetId then bypasses the proposed-parent guard, creates a human-audience backlog child and attaches the existing session. This path remains at v0.3.0 source and was not exercised live. Concurrent-edit handling remains unknown.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Artifacts can retain a plan or requirements document with a thread; a dedicated executable requirements schema is not established. Workspace users allowed to open the thread can edit its artifacts. Every selectable agent mode can read and write artifacts using the artifact tool. Every artifact save creates a new revision. Artifacts remain accessible from the thread Artifacts view until deleted.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-07. Scope and sources
Podium
Sourcedpartial
The spec CLI and service (living specs under pspec with stable SP IDs) are in v0.3.0. Specs is an experimental, default-off feature listed on the edge channel: the flag controls the Specs UI, the spec line in CLI help and whether agents are told about the spec CLI. The podium spec command itself is dispatched, and the server specs service has no flag check, so direct CLI use does not need the flag. This agrees with the docs, which call specs not baseline rather than blocked. Not run end to end.
The Orb guide describes several independent tasks running at once, each with its own environment. Several independent orb threads can run at once, each in its own orb. Each user gets a burst of 20 metered orbs, then one new orb every five minutes; higher limits through support; Enterprise sets per-user spend entitlements. Activity filters threads by status; Space cards show live status; desktop tray menus mark running, unread, waiting and failed threads.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Docs and release notes describe many sessions running as independent terminals, one worktree per issue, and several paired machines under one server, with session and issue state visible on the flight deck, board and CLI. Agent-initiated spawns are limited to 10 per target issue per UTC day. Achieved concurrency is unmeasured.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
@Amp in a public Slack channel linked to a workspace-shared Space starts work there. Puck and Spaces start agent threads; Linear issues can be assigned or delegated to Amp, with agent routing to projects. Tagging a workspace member invites them and turns on multiplayer; assigning a thread to a human owner is not documented. Threads and Spaces have one owner: only the owner switches a thread's model, a Space's owner controls its settings, and Space transfer needs acceptance. Exclusive work claims are not documented. Orb starts beyond the pacing limit wait in line; messages can be queued until the agent finishes; webhook events queue per endpoint. A task queue for dispatching work is not documented in the checked pages (Orbs, Threads, Automations, Spaces).
Checked 2026-11-12 (10 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Issues carry an assignee, claim sets assignee and in_progress, a coordinator session can be set per issue, sessions have seats, and work can be placed on a chosen machine or agent profile. claim has no exclusivity check of its own, locks are advisory leases and Podium does not serialize agents editing one worktree, so exclusive ownership is not enforced.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Task executes delegated work and Read Thread summarizes threads. Built-in Task subagents follow the thread's model and effort; custom plugin agents can define model choices. Further scoped facts are recorded in the atomic answers. Amp uses subagents automatically for suitable tasks or when asked. The main agent can give a Task subagent another task with its earlier context; delegated threads can receive more instructions. By prompt, through Puck, from a Space, or by an agent starting another thread. The main agent receives subagent summaries only and users cannot guide subagents directly; delegated threads and Space cards are visible live. The main agent for subagents, Puck, or a Space. Subagent steps stay in the thread and delegated threads persist as their own threads.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium orchestrates delegation itself: podium agent spawn starts a full child session on an issue with a chosen harness, model, effort or agent profile, mail carries messages, and bounded await reports the child result. Children are real sessions shown as seats on the flight deck. Agent-initiated spawns take a daily budget per issue. A complete delegated task with retained child findings was not run.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Amp saves one schedule per thread; it can run once or repeat, with pause, resume and delete controls. GitHub events can wake existing threads or start new Orbs. Further scoped facts are recorded in the atomic answers. Amp stores the schedule on the thread, wakes the agent and resumes the orb when it fires. Schedules repeat until paused, deleted, ended or a completion condition is met; one schedule per thread. Failed runs show the error; temporary delivery errors can keep a recurring schedule enabled, and Amp may pause it. Plugin webhook events retry with backoff up to 5 minutes and are dropped after 1 hour.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
The server owns a scheduler for cron and one-time automations that spawn agent sessions, skipping overlaps and not backfilling occurrences more than an hour late. The Automations UI is an edge-visibility feature that defaults off and docs list automations as non-baseline. Agents can subscribe to issue and session events.
A snoozed thread surfaces and notifies when it needs an approval or answer. The same thread opens in CLI, web and apps; the orb Terminal tab is a shared tmux session on the agent's working copy and can be embedded in a reply. A scheduled run can unsnooze a thread; snoozed threads waiting for approval reappear and notify. Archiving pauses an orb and automations can be paused; snoozing does not stop the agent. Pausing a running turn is not documented. A paused orb wakes on the next message with conversation, files and services in place. Threads keep prompts, replies, tool calls and file changes. Messages sent while the agent works steer it after the current step; Task subagents receive new tasks as steering. Pressing Esc twice interrupts the agent; archiving stops the orb before it pauses on its own; thread history remains. Failed automation runs show the error; failed repository setup links a Setup Log; tray menus mark failed threads.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Chat and native CLI views share a session. v0.3.0 defaults human controller chat to steer and immediately types it into running Claude Code, Codex, Grok and OpenCode TUIs that declare native busy-input queuing; undeclared harnesses wait for a boundary, and compaction waits. Mail and automation keep when-ready delivery. Stop, resume, questions, decision issues and offers remain available. This is inspected implementation, not a cross-harness runtime parity result.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Unread dots, taskbar or launcher counts, and tray dots for waiting threads; apps notify when a thread is ready. Activity lists every visible thread with time, status, author and project filters, including archived and snoozed. Child threads nest under their parent in the sidebar and move with it; parents can archive all children at once.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Release 0.3.0 ships a flight deck showing a whole mission in one view: phases, every task and its state, who works on what, what is blocked and by what, and proposals. Done work stays visible. The CLI prints the same snapshot. The waterfall source is dormant and the Timeline view was deleted, so per-task timeline history is not shown. The shared 12-task UX scenario has not been run.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-11. Scope and sources
The model selector sets each thread's model, effort and speed, replacing the Dial removed on 2026-11-12; plugins can define custom agents. Spaces start a thread per piece of work and chart shipped, archived and open threads. The thread owner can switch model and effort at any time, and the transcript records the switch. Automations, Puck and Spaces take plain-language instructions. Presets include saved prompts (Ship, review, custom), plugin-defined agent modes and Space thread models. amp -x runs one prompt; a new thread starts from the sidebar. Nested child threads stay grouped with their parent; artifacts link back to the thread that wrote them.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
A quick task can be an interactive session or an issue; an issue can grow into a mission by gaining children while keeping its identity; no mandatory workflow preset applies in the baseline. Per-agent harness, model and effort are chosen at spawn or through agent profiles, and the deck shows seats. The coordinator-with-two-reviewers scenario was not run.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
The CLI can run Amp on the user's own machine. Users can supply their own Orb compute through amp orb provider and a dispatch program. Orb execution can be hosted on user infrastructure; this is not self-hosting the Amp control plane. Orbs are per-thread machines that continue working when the laptop closes. The size table lists per-Orb CPU, memory and 60 GB disk; default-size rules vary by project and plan. An orb agent keeps working with the laptop closed. Orbs pause after idle timeouts, not run limits; paused filesystem and memory stay encrypted until the thread is deleted. Per thread: Amp orb, a user runner, a self-hosted orb provider or a local CLI session. Open the thread from any app, the web or the CLI and keep prompting. Amp-hosted orbs bill the thread owner per minute from tier minutes or credits; runners are free; self-hosted orbs are free except on Enterprise. Amp provisions an orb per thread from the project snapshot automatically. Primary infrastructure is GCP in the US; Enterprise lists regional endpoints for BYOK providers. No orb region choice is documented.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
The published Podium execution choices are local user machines and a user-owned VPS or paired machine. No vendor-operated agent host is in the complete download offering.
Runner worktree creation also creates a named branch. A runner creates a Git worktree and branch for a new thread; worktrees separate files without asserting a security sandbox. A documented Docker provider example runs each Orb as a container; this is a self-hosted execution option. Each Orb thread has a fresh isolated working environment; the shared terminal uses the same filesystem as that thread's agent. Docker-backed Orbs do not preserve memory across pause; terminals and desktop processes are lost on wake. Each runner worktree has its own working files and branch. Each orb thread gets its own fresh environment and working copy cloned from the project. Orbs reach the internet and can route web traffic through a user runner; Enterprise lists network and MCP access controls. Egress restrictions inside orbs are not documented. Secrets are injected from settings, OIDC issues short-lived credentials, GitHub tokens are fetched short-lived per command, and redaction keeps detected secrets out of threads. In multiplayer, every contributor can access the orb, its secrets, files and terminal; no hostile-code containment claim is made. Orbs are sandboxed cloud machines on e2b compute; the isolation technology is not specified.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Podium separates issue files with Git branches and worktrees, but its released daemon launches native host processes without product-managed container or VM provisioning.
Docker-backed Orb services restart when a portal is reopened after wake; terminal and desktop process state is not restored. An idle Orb wakes with conversation, files and services retained; worker-crash recovery was not tested. Docker-backed Orbs retain disk but lose terminal and desktop process state when paused. The orb agent keeps working when the client disconnects. No sandbox expiry: paused orbs are kept until the thread is deleted. The orb Terminal tab gives a shell on the agent's working copy. A sleeping orb costs nothing even for weeks and wakes with conversation, files and services. Services pause with the orb and are available again when it wakes; .agents/resume reconnects after each wake. Project snapshots plus .agents/setup and .agents/resume rebuild and repair orb state. Stored threads open in any client.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Closing a browser tab does not stop the agent process on a running Podium host. Idle sessions may hibernate under configurable policies; a stopped session keeps its transcript, branch and resume reference. Release 0.2.0 says a failed update rolls back and running sessions keep running through the upgrade; database rollback remains a manual restore. Terminal reattachment goes through the driver adopt contract with process-identity and observation-lease checks. Worker-crash and multi-day behavior were not tested.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Install shared dependencies in .agents/setup. A matching prepared project snapshot is reused for up to 72 hours. Snapshots capture repository and setup changes and can be deleted from settings or CLI. Long-running processes belong in .amp/services.yaml or systemd, not setup. Shared setup and runtime use different credentials; setup credentials are removed before snapshotting. A matching project snapshot skips setup; freshness is limited to 72 hours and supported Git host checks. .agents/resume runs after activation and wake for authentication, repair or reconnect; successful automatic repair is not independently tested. .amp/services.yaml declares services, Amp manages PORT, and portals expose orb web servers. Pre-clone, pre-setup, .agents/setup, snapshots and .agents/resume prepare remote environments.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Only machine setup is documented: the installer installs missing runtime tools and pairs a host, and the state directory holds server-managed secrets. Dependency caching, services, ports and automatic repair for project environments were not found.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Setup stops after 20 minutes; a failed or timed-out setup does not publish a refreshed snapshot. Amp checks spend against each user's effective Enterprise entitlement; whether a reached limit stops work is not stated. Enterprise per-user entitlements such as $100/day or $50/week; Enterprise lists spend limits and reporting. No run-time limit: orbs pause only after 5- and 20-minute idle timeouts. Plugin webhook handlers have 30 seconds.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium exposes retryable errors, optional auto-continue with backoff, bounded waits and provider quota. Claude stream and hosted headless drivers enforce a supplied turn deadline, with no deadline when timeoutMs is absent. The one-shot headless server runner supplies its budget, default ten minutes. Agent spawns have a daily per-issue budget. Operator-configurable task-wide turn, token and spend caps remain unestablished; quota display is not enforcement.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
A Space chart counts the last 30 days of threads as Shipped, Archived or Open (Done where commits cannot be seen). Threads waiting for approval surface in the sidebar, even when snoozed, and notify. Failed automation runs show the error and failed repository setup links a Setup Log. Setup logs in orbs; threads keep every tool call and result. Each thread shows its cost so far. Thread storage preserves prompts, responses, tool calls, tool results and attachments as an audit trail. Usage Breakdown exports CSV; amp usage shows balance; a data management API exposes analytics.
Checked 2026-11-12 (10 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium offers local logs and crash export, structured session status, issue stats, a usage and analytics sheet with hourly list-price cost, per-task costs and host load panels. The 0.3.0 notes report a vendor benchmark of about 330 MB server memory against about 1.2 GB for an earlier Podium build, and 8 to 52 percent faster start; the method is published in the repository and was not reproduced here. Usage export and cross-reconnect consistency were not established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-11. Scope and sources
A thread's dollar indicator opens detailed usage; this is documented, not reconciled against invoices. Orb minutes are priced per size and shown separately from model spend. Usage Breakdown exports CSV totals per period and a member summary. The $ price in each thread opens its cost detail. Tier allowances cover included usage; own subscriptions carry no Amp token fee and do not count toward entitlements. Per-thread dollar cost is shown in the thread. Usage can be viewed as tokens, and Amp explains a thread's token use on request.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Since 0.2.0 Podium shows what each task costs, on the task header and the usage sheet, with per-session rows and descendant rollups. The 0.3.0 deck adds a mission cost chip. Figures are token estimates at public model list prices, not subscription invoices or provider bills.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
After a burst of twenty metered Orb starts per user, later starts queue at one every five minutes. Orb starts beyond the pacing limit wait with an estimated wait instead of failing. Archiving stops an orb; cancelling a waiting orb start is not documented. A 20-orb burst per user, then one new orb every five minutes. Orb size sets CPU and RAM per thread. Pacing is per user; Enterprise lists higher resource quotas and rate limits without numbers.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium reports host memory, load and disk, auto-hibernates idle sessions under pressure, and provides named advisory resource leases. The merge-queue panel was removed; a shipping queue remains behind the Shipping flag. Concurrency limits and bounded test queuing were not established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Workspace members can chat with each other in the thread. Workspace teammates can prompt together in an enabled multiplayer thread. Further scoped facts are recorded in the atomic answers. Orb multiplayer does not need the owner online; runner multiplayer ends when the owner stops sharing the runner. Multiplayer needs a non-private thread in an orb or shared runner, lasts 7 days, and bills the owner. Restricted, workspace or link sharing; Enterprise adds groups.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The issue UI displays owner and visibility read-only and deliberately omits sharing controls. Backend share handlers and a Members settings section with invites and admin or member roles exist; they do not establish a finished human collaboration workflow. Docs say small teams can share an instance but the interface prioritizes one operator.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Member and Admin roles are documented; Admin manages membership, billing, settings and policies and can view private threads. Admins invite members; member email-invitation rights are configurable. Teams list SAML/OIDC SSO; Enterprise adds SCIM. Enterprise supports directory sync of members and groups. A non-enterprise workspace needs a paid Megawatt or Gigawatt member for SAML/OIDC SSO; Enterprise does not require a paid member tier. Workspace admins and the project creator edit project settings; other project roles are not documented. SCIM deprovisions users; admins remove members and can disable service accounts. Thread and Space Contributor or Viewer roles. Admins open restricted threads only through passkey re-authentication with audit logging; only workspace members can be added.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Podium defines member/admin instance roles and admin-only membership management. The released client authentication uses password-backed sessions and paired-device tokens without enterprise SSO.
Multiplayer starts with a seven-day duration, can end early and expires automatically. Stopping runner sharing ends multiplayer on threads created on it. tool.call hooks return allow; amp.mcpPermissions rules allow servers. In multiplayer, the owner or a workspace admin approves box and secret changes and the owner approves Slack posts; contributors cannot. No approval by default; a policy plugin can gate commands. By default Amp runs tools without asking. Ship is a user action; Slack posting asks first by default; making a portal public requires approval when the agent asks. tool.call returns reject-and-continue; MCP rules reject servers; amp.tools.disable removes tools. The tool.call hook runs before every tool call. tool.call hooks, amp.tools.disable and amp.mcpPermissions. Portals are limited to thread viewers; public access lasts 1 hour to 7 days and can end early; runner portals cannot be public.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Agent-session management requests go through an operator approval broker with approve and deny; the privacy docs say lifecycle commands and automation requests may require operator approval. Harness permission prompts appear as a Permission needed state; credential copying during pairing is a per-grant opt-in. Command-level policy enforcement beyond the harness own modes was not established.
Enterprise admins can access authentication event audit logs; comprehensive application audit logs remain internal. Each artifact save creates a revision and Share identifies the last editing thread agent. Artifacts export to Markdown, HTML or PDF; threads export to Markdown or owner-only JSON. Artifacts are editable, with a revision made for each save. Deleting a thread deletes its artifacts. A stale agent artifact write is refused when another revision was saved; the agent must merge its changes. Threads are stored on Amp Server and reopen across sessions. Enterprise threads and author attribution remain when users leave; non-Enterprise threads go with the account.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Sourcedpartial
Issue actions, approval steps, comments and mail are recorded server-side. Issue activity events, including approval steps, are pruned by the server's janitor once they are older than 14 days or fall outside the newest 50,000 rows. That policy does not define all content retention, export or tamper-evident audit guarantees.
The Changes pane offers all changes, uncommitted changes or a selected commit. The commit picker supports the right arrow key to read a commit message. Commits carry an Amp-Thread-ID trailer linking a code change back to its conversation. The portal review widget sends element comments to the agent; Space messages take comments. The Changes pane orders files by importance and picks commits; portals and Desktop show the running app. The phone apps give the same threads, orbs and Puck as ampcode.com, plus notifications, dictation and sharing from other apps; review-specific narrow-screen guidance is not documented in the checked Apps and iPhone pages. Reviewers reply in the thread or comment on the portal; a separate request-changes state is not documented in the checked Threads and Portals pages.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Review is a documented loop: issue stage review, diff and checks, artifacts, offers with merge or send-back buttons, and a Tray of review-ready work. Release 0.3.0 adds quoting text from a reply or review finding into the next message and a full-height reading view with an outline. The board has a key map with a help sheet. Guided-review quality and narrow-screen use were not measured.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Colleagues use the same thread and Orb or shared runner while multiplayer is active; disconnect takeover was not tested. Teammates and agents edit thread artifacts with revisions. Multiplayer contributors prompt the same thread. Transfer of thread ownership is not documented in the checked pages (Multiplayer, Threads, Spaces), which document only Space ownership transfer. Orb threads continue without the owner; runner multiplayer ends when the owner stops sharing. Contributors get the orb, secrets, files and terminal; the agent refuses account changes in multiplayer. Contributors use the orb's terminal, files and portals. The full thread history is shared.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The durable issue, session transcript, artifacts and mail persist for any later agent, and a live session can be handed to another machine in an edge feature. Colleague takeover by a second person was not established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
The thread's Changes view shows the diff, alongside files, services and conversation. Thread artifacts retain plans, reports and notes. The Artifacts tab groups multiple documents in folders. Artifact links in the transcript open the document in its thread. Artifacts and their uploaded images remain with the thread; deleting the thread deletes its artifacts. A child thread can edit or copy an accessible parent artifact, then notify its thread of the change. Portals expose orb web servers for preview. Desktop runs report what they did with screenshots, and delegated threads can return screenshots when asked; no screenshot gallery is documented. The thread holds the conversation, changes, running services and proof; test results are not a separate retained artifact.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Artifact commands store immutable server-side copies of issue-linked files, including images, video, HTML and documents, which render in the issue sidebar and offers and stay readable after the worktree is gone. Each task has an artifacts page grouped into screens, documents and reports, with filters and search. Revision linking remains untested.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-07. Scope and sources
Default Ship asks the agent to run the full test suite. The documented agent instruction ignores a failing test only after confirming that it fails on the base branch too. This describes a prompt, not an independently enforced gate. A project can configure Ship Behavior or write a Custom Ship prompt. CI subscriptions follow an exact commit's checks and report failures. The default Ship prompt runs the full test suite before pushing, but gates are prompt instructions, not enforced policy. Optional: the Task subagent review skill, the oracle plugin and a built-in review prompt.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
In the experimental Shipping path a configured validation profile runs before landing (default argv bun run test with a 10 minute timeout) and a failed validation holds the order. Lint, type, CI and spec conformance gates and independent review requirements were not found as product-enforced policy.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Push to Branch ships to a feature branch instead of the base branch. Default Ship commits uncommitted files before rebasing and pushing. Orbs can push branches and open PRs using a connected GitHub identity. Default Ship rebases on the latest base branch and repeats after a concurrent push. The agent resolves conflicts and asks the user before proceeding on substantive conflicts. A PR linked by thread URL or ID appears in the thread title bar with open, draft, queued, merged or closed status. The agent follows CI for a commit and addresses PR feedback from subscriptions. PR subscriptions deliver reviews and inline review comments to the thread. Runners create a Git worktree per thread when asked, also for agent-started threads.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Issues own branches and worktrees, and issue actions rebase, open a pull request with gh (pr create --fill) and merge, subject to a merge-style setting (ff-only by default, with auto-rebase before merge). Podium does not resolve branch conflicts, and PR review comments and CI response were not established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Default Ship writes directly to the base branch after rebasing; PR approval and merge-policy enforcement are not established. Ship behavior is a project-configured agent prompt. The source does not establish an immutable authorization boundary. A user presses Ship to start delivery, and substantive rebase conflicts are checked with the user; a separate approval gate is not documented in the checked Shipping page. amp boxes deploy builds and serves an app on a box; service-account CI can redeploy.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The operator accepts work through offers and runs the merge action under a branch merge-lock; in the experimental Shipping path approval is frozen to an exact head and delivery lands on a local target branch by default. Release and deployment are not product features found in the sources.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
The thread records linked PRs when the PR description includes its thread URL or ID. Threads show commit and PR previews with status; CI subscriptions report check results. Commits carry an Amp-Thread-ID trailer, and a PR description with the trailer lists the PR in the thread. Space progress counts a thread shipped when its commits reach the default branch, its PR merges or Ship archives it. Box deploys return the app URL on onamp.dev; linking a deployment to the thread is not documented. The thread keeps conversation, code changes, running services and proof.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Shipping orders record approved base and head SHAs, tested integration SHA, landed ref and validation result in a durable delivery receipt, and stale approvals hold the order. Shipping is experimental. The pr action records a PR URL on the issue. Build and deployment links and rollback were not found.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
TypeSafe's Jev model suggests the next action in a thread; no review queue is documented.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The shipping service records revision-specific approval, configured validation and delivery receipts. The Shipping panel is an experimental edge toggle, off by default; the separate Queues panel was removed.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-07. Scope and sources
Amp access is covered by the limited, revocable service license in its custom terms. The hosted service is governed by Amp License Terms; an open-source server license was not established. Plugins and skills are user-supplied code in personal, workspace or project locations; licenses of Amp's own published plugins were not established.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
The public repository is licensed Apache 2.0, and the website presents Podium as an open-source agent development environment. Hosted extensions and hosted-service terms are not established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Amp explicitly says it does not offer self-hosted deployments of its control plane. Self-hosted Orbs are execution workers. Threads export to Markdown, with full JSON available to the creator. Thread export formats are Markdown and JSON; artifacts export Markdown, HTML and PDF. Thread transcripts can be exported; importing an exported history was not established. An Amp account is required and the server is not self-hostable; self-hosted orb providers still connect to Amp. What remains usable without Amp is not documented in the checked pages (Security reference, Self-hosted orbs). Cancel any time; the paid tier lasts to the period end. Account deletion removes owned threads within 30 days, except Enterprise workspace threads. Amp Server handles accounts, threads, sync and usage; model requests go from Amp servers; e2b provides orb compute.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Local desktop and self-hosted installs run on user-controlled machines with a documented state directory. The docs describe stopped full-state backup and automatic pre-migration database-only snapshots retaining three copies. Those snapshots omit artifacts, transcripts, uploads, configuration and secrets; no portable structured export/import contract was established.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-11. Scope and sources
Local code snippets and context can be sent to Amp Server and model providers. Prompts and conversation history are sent to Amp Server and then inference providers. Authentication audit logs are available to Enterprise admins; internal application logs are not exposed to admins. Enterprise workspace threads remain until the enterprise contract ends, including after the author leaves. Deleting a thread also deletes its associated Orb, usually within seconds. Training can be toggled in Settings Privacy; workspace enabling requires admin approval. Amp Server stores data in a multi-tenant Google Cloud Platform project; its primary infrastructure is hosted in the US, and no narrower region is stated. Personal provider subscription terms and data controls apply independently of Amp training settings. Attachments stay private to the thread under access checks; deleted threads and their data are removed within 30 days. Training requires explicit opt-in and workspace admin approval, never on Enterprise; no first-run consent dialog is documented. Best-effort secret redaction before content reaches threads or providers; stored secrets are encrypted; the CLI keeps credentials in ~/.local/share/amp/secrets.json. The security reference lists GCP, Anthropic, OpenAI, SpaceXAI, Meta, Vertex, Bedrock, Baseten, TypeSafe, e2b, Pierre, LiveKit, Parallel, Cloudflare Email, WorkOS, Stripe and others, with what each sees. Telemetry is stored with thread and user data in a GCP project; what is collected and opt-out controls are not described.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Docs state repositories and native harness credentials stay on machines you control and server state lives in a documented directory. Since 0.2.0 a daily usage report is on by default and can be turned off; crash reports stay opt-in. Usage reports carry an install id, schema version, Podium version, OS, architecture, bucketed install age and machine count, session counts per harness and whether issues were used. Crash reports carry the same install id, versions, OS and architecture, a fixed error type and up to 20 stack frames, each with a Podium source file path, line and function when there is one; error messages and frames outside Podium are dropped. Neither tier sends prompts, transcripts, your code or paths outside the Podium source tree. Model-provider handling is the harness own. Hosted analytics and website analytics are separate paths.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-21. Scope and sources
Public tiers are Hobby free, Megawatt $20/month, Gigawatt $200/month, Teams at no additional workspace charge and custom Enterprise. Team members retain individual tiers; model/tool and Orb usage are separately metered. No seat minimum: workspaces cost nothing and members can stay on Hobby; SAML/OIDC SSO needs one paid member; Enterprise has pooled credits and no paid seats. Invoices are issued through Stripe, which accepts VAT IDs; tax treatment is not otherwise stated.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
No Podium price is published: podium.do/pricing returns 404 and the docs list billing as not implemented. The software is open source and free to download; model and compute costs are the user's own.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Orbs cost $0.08-$2.13 per hour by size, billed per minute and only while running; Megawatt includes 45,000 orb minutes. Hobby $0, Megawatt $20 ($10 with the education discount), Megawatt 50 $50 and Gigawatt $200 per month; Teams adds no fee; Enterprise is custom. Gigawatt is listed in the docs pricing table (last modified 2026-11-17) and was recorded at $200 on 2026-11-13; the interactive pricing page shows it only behind a toggled tab, so it was not re-observed in that page's static HTML on 2026-11-21. Megawatt includes $20 of agent usage and 45,000 a1.small orb minutes; Megawatt 50 includes 2.5 times that; Gigawatt includes $200 of agent usage and 1,000 hours of a1.xxlarge orbs. Included usage resets monthly without rollover. One person: Hobby $0 with pay-as-you-go credits and orbs, or Megawatt $20 with included usage; model usage beyond is billed at provider API price. Two people: a free Teams workspace, each member on Hobby ($0), Megawatt ($20), Megawatt 50 ($50) or Gigawatt ($200); by default the workspace pays its members' paid tiers, and multiplayer costs go to the thread owner. Megawatt-only scenario: five people cost $0 (all Hobby) to $100 per month (all on Megawatt) in platform tiers; the range excludes Megawatt 50 ($50 each) and Gigawatt ($200 each). A Team workspace can set a Megawatt or Gigawatt base tier for all members and by default pays members' tiers. Credits and orb minutes beyond included usage are extra; workload cost is not priced. Credits pay actual model and tool usage without markup for non-enterprise accounts; orbs bill per minute by size; boxes bill separately. Per-task model spend varies and is not estimated in advance; Amp can explain a thread's usage afterwards. Enterprise pricing is custom.
Checked 2026-11-21 (1 day old); agent cross-checked 2026-11-21. Scope and sources
Podium
No sourced public findingnot-published
No publishable sourced finding is available for this criterion. This does not mean the product lacks the capability.
Dated releases include Android (Oct 27), Artifacts (Nov 2), Service Accounts (Nov 3), GitLab (Nov 5), Spaces (Nov 6) and Boxes (Nov 10, 2026). macOS 26+ on Apple silicon; iOS 26+ via TestFlight; Android 13+ by APK; Windows 10 1809+; listed Linux distributions; a1.3xlarge only for Gigawatt and Enterprise. The JetBrains integration is deprecated without updates; Amp states it removes unused features and keeps no backward compatibility. Threads export as Markdown, or full JSON for the creator, from the menu or CLI. The iOS app is a TestFlight beta pending App Store release. ampcodestatus.com with email, RSS and Slack alerts; @ampcodestatus on X. Bug reports from clients, help via X or email; Enterprise lists priority support.
Checked 2026-11-12 (10 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium is pre-1.0. Stable v0.3.1 (2026-11-06), an update fix, follows v0.3.0 (2026-11-05), v0.2.0 (2026-10-11) and v0.1.0 (2026-08-21), with edge and dev prerelease channels. Windows and Linux desktop are Experimental, the native iPhone app is in private beta and hosted Podium is unconfirmed. Support runs through GitHub issues, Discord and private vulnerability reporting.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Amp publicly states SOC 2 Type II certification; the underlying report was not inspected. Vendor security statement only; certificate issuer, validity period and report scope were not independently verified. SOC 2 Type II reports are available on request through the Security Trust Portal; not publicly downloadable. The vendor SOC 2 Type II statement was re-read on 2026-11-21; the report, audit period and covered services were not accessed.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
No sourced public findingnot-published
No publishable sourced finding is available for this criterion. This does not mean the product lacks the capability.
The Hobby tier is recurring free; model usage and hosted compute may still cost money. Self-hosted Orbs and own subscriptions or API keys have no Amp fee for individuals and teams; infrastructure and provider costs remain. Hobby includes no usage. Own subscriptions and API keys have no Amp token fees or limits outside Enterprise; provider limits still apply. The iOS app is a public TestFlight beta that needs no invitation. Hobby is free with own subscriptions or keys, free self-hosted orbs and runners, and pay-as-you-go orbs; a Teams workspace with sharing and multiplayer costs nothing extra. a1.3xlarge needs Gigawatt or Enterprise; SAML/OIDC SSO needs a paid member; service accounts need Enterprise or bought Team credits; SCIM, audit logs and IP allowlisting are Enterprise.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
Agent cross-checkedpartial
Podium is open-source Apache-2.0 software, and the download page calls it free and open source. Download links for macOS, Linux and Windows redirect to GitHub release assets without a login (observed by HTTP requests; not installed). No Podium charge is documented for self-hosting; agent provider usage is separate. The native iPhone app is private beta and hosted access is unconfirmed.
Checked 2026-11-06 (16 days old); agent cross-checked 2026-11-06. Scope and sources
Agent usage is billed in credits and Orb compute by the minute. Megawatt includes $20 agent usage and 45,000 minutes of a1.small Orbs monthly. Paid credits permit continued model and tool usage beyond the monthly allowance. Self-hosted orbs are free on Hobby, Individual and Teams; Enterprise workspaces pay 100% of the standard orb price, billed per minute while the orb is connected to a thread, with volume discounts on request. The guide states no provider API markup for individuals and non-enterprise workspaces. Paid credits require no paid tier or commitment. Enterprise uses custom pricing. Cancel any time in user settings; the tier stays active to the end of the month. Partially specified: tier prices including Gigawatt $200, orb rates and no-markup credits are public; Enterprise is quote-only.
Checked 2026-11-07 (15 days old); agent cross-checked 2026-11-21. Scope and sources
Podium
No sourced public findingnot-published
No publishable sourced finding is available for this criterion. This does not mean the product lacks the capability.