v0.4.3

Agents v0.4.3 coordinates AI coding agents through task state, handoffs,
findings, verification, and human approval.

See README.md and tutorials/tiny-todo-cli.md in each archive for a
quick start.

Changelog

• c3552b9 Add Gemini and Antigravity providers

v0.4.2

Agents v0.4.2 coordinates AI coding agents through task state, handoffs,
findings, verification, and human approval.

See README.md and tutorials/tiny-todo-cli.md in each archive for a
quick start.

Changelog

• e3a7bee Add durable task plan mode
• 4d3f425 Add plan review gate
• 8007ad8 Archive plan refine task
• 29b6637 Harden reviewer branch parity checks
• 577c958 Improve archive summary hygiene
• 1eae2e1 Make plan drafts free-form
• e34b9c3 Make plan mode substantive
• ad1b317 Replace plan revise with refine

v0.4.1

Agents v0.4.1 coordinates AI coding agents through task state, handoffs,
findings, verification, and human approval.

See README.md and tutorials/tiny-todo-cli.md in each archive for a
quick start.

Changelog

• c023625 Update release workflows for Node 24

v0.4.0

Agents v0.4.0 coordinates AI coding agents through task state, handoffs,
findings, verification, and human approval.

See README.md and tutorials/tiny-todo-cli.md in each archive for a
quick start.

Changelog

• 016c8a5 Add Agents MCP server
• ef11495 Add MCP framing coverage
• 973f13c Add cleaned run activity watch
• 379878f Add local AI coordinator advisory
• 59e0c4f Add local AI first-pass reviewer
• cb766c1 Add optional local AI coordinator classifier
• 3dde38e Add provider map configuration
• edd5fd9 Archive provider map task
• 2c47dcf Archive reviewer pipeline task
• 8ade5ed Clean Cursor stream output and allow Codex greenfield reviews
• 301a731 Coalesce split run activity sentences
• 3e0d410 Ignore no-op local reviewer findings
• 81768e5 Improve go continuation loop UX
• 91f6476 Improve interactive coordinator flow
• 8230727 Keep quiet agent runs visible
• 9f2e8e8 Move review config after roles
• 3682f4f Render watch activity as markdown
• 062d566 Report archive preservation counts
• 905454e Show human approval boundary in supervision
• cb56ef6 Show observed progress during quiet runs
• f83629d Skip local AI tests when loopback is unavailable
• 8029f1f Stream Cursor Agent output events
• 484c7de Structure run activity recording
• 33f0b24 Support reviewer role pipelines
• d46c0a2 Use structured local reviewer results

v0.3.0

Agents v0.3.0 coordinates AI coding agents through task state, handoffs,
findings, verification, and human approval.

See README.md and tutorials/tiny-todo-cli.md in each archive for a
quick start.

Changelog

• bb23082 Add GoReleaser config for v0.3.0 release
• d7e0c76 Add license before public release
• ae5fbec Ignore placeholder provider directives
• af4ae89 Improve GoReleaser packaging
• eaacbed Keep planning chat in coordinator
• eee3251 Render interactive Markdown replies

v0.2.2

Changelog

• 6187975 Add PTY coverage for unified loop view
• fdf25fb Add structured role progress
• a7ca648 Archive unified-loop-view task: M0-M3 delivered, 3 findings verified, decisions folded into ledger
• 0c1e0e1 Clean interactive role output
• 9d51b93 M0 loop events: emit lifecycle events from the managed loop, add agents run --json-events
• be1ec43 M1 agents go: one-command entry point from intent to a running loop
• 34f8c3a M2 unified split view: two live lanes and a status bar for agents go
• af46ed7 M3 Ready-for-you screen: human-boundary summary with accept, iterate, abandon
• 6594b1f Open unified-loop-view task: decision and implementer handoff (plan in local qa/)

v0.2.1

Changelog

• 6500469 Harden prod edge cases: trust-gate run, cap slugs, resolve root from subdirs, flag dangling pointers
• 82b80d8 codex driver: re-add .agents to sandbox writable roots
• 39abd1e init: ignore ephemeral .agents artifacts (tmp/, active); steer agents off in-repo caches

v0.2.0

Changelog

• 0acd889 Archive verification-loop task: 4 findings verified by codex, evidence recorded
• 104d165 Enforce actor identity and independent verification in the run loop
• a54b3e7 Handle interrupts during managed role runs
• 9d10dc6 Harden verification evidence: task check, timeout, stamp ordering, JSON parity
• 0449e69 Qualify records with the agent model and add task chunk setters
• ea2c471 Serialize progress writer output

v0.1.1

Changelog

• aba404a Fix managed agent process shutdown

v0.1.0

Changelog

• 6abe60f Add interactive agent loop and release tooling
• 833965f Initial Agents commit