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