Skip to content

Agent Missions MVP: settings, missions API, tracker actions, Telegram /mission#2

Draft
computerbox124 wants to merge 2 commits intodevelopfrom
cursor/agent-missions-mvp-3563
Draft

Agent Missions MVP: settings, missions API, tracker actions, Telegram /mission#2
computerbox124 wants to merge 2 commits intodevelopfrom
cursor/agent-missions-mvp-3563

Conversation

@computerbox124
Copy link
Copy Markdown
Member

@computerbox124 computerbox124 commented Mar 29, 2026

Summary

Implements the Agent Missions MVP as a thin extension on existing pod-ai-bot (Mongo config DB), ai-bot REST types, ai-bot-resources UI, model-ai-bot settings category, tracker actions, and pod-telegram-bot /mission command.

Backend (pod-ai-bot)

  • Mongo collections: agentProfiles, executorResources, missions
  • REST routes: /agent-profiles, /executor-resources, /missions, /mission-channels/status (Bearer workspace token)
  • Executor resolution: explicit → profile default → first non-busy shared executor of requested type (executorType on create when no default)
  • Allowlisted spawn: argv from MISSION_CODEX_COMMAND / MISSION_CURSOR_COMMAND (defaults to cat for demo), stdin prompt, timeout, output cap
  • Context: tracker issues (classic bucket todo/backlog), max 20 tasks; collaborative descriptions stubbed in MVP
  • Results: chat comment on target issue (task scope or first issue in list for project/bucket)
  • Agent profile lookup by id, name, or role (Telegram agent=CTO)

Client

  • Settings → Agent Missions (agents, executors, channels, recent missions)
  • Run Agent Mission popup from tracker issue/project actions

Telegram (pod-telegram-bot)

  • /mission with missionParse.ts + mission.ts calling POST {AI_BOT_URL}/missions
  • Multi-workspace: require workspace= when multiple integrations exist

Automated QA (unit tests)

  • pod-ai-bot src/missions: output normalization, executor resolution (explicit, default, shared fallback, private access, busy shared), filterExecutorEnv (mocked config)
  • pod-telegram-bot missionParse: basic parse, workspace=, task= + scope=task, invalid scope

Manual / E2E

Full Huly stack (accounts, transactor, Mongo, browser login) required for UI E2E; not run in CI agent VM.

Tradeoffs / deferred

  • Task descriptions: placeholder text (collaborative markup not loaded in MVP)
  • Default executor binary: cat unless env overrides to real CLI tools
  • Full postgres integration tests in telegram-bot package still require DB (unchanged)
Open in Web Open in Cursor 

cursoragent and others added 2 commits March 29, 2026 22:08
- Extend ai-bot REST DTOs and pod-ai-bot: agent profiles, executors, missions CRUD
- Allowlisted executor spawn (default cat; MISSION_CODEX_COMMAND / MISSION_CURSOR_COMMAND)
- Mission context from tracker (classic bucket=todo|backlog), cap 20 tasks
- Post results as issue comments; resolve agent by id/name/role
- Settings Agent Missions + Run Agent Mission popup from tracker actions
- Telegram: /mission parser, AI_BOT_URL to pod POST /missions, multi-workspace hint
- Tests: output normalization, mission command parsing

Co-authored-by: Huseyn Hajiyev <computerbox124@users.noreply.github.com>
…d env filter

- Extract resolveMissionExecutor for pure unit tests
- Cover explicit/default/shared fallback, private access, busy shared
- Add filterExecutorEnv tests with mocked config
- Expand Telegram missionParse tests (workspace, task, invalid scope)
- gitignore qa-artifacts for local recordings

Co-authored-by: Huseyn Hajiyev <computerbox124@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants