Skip to content

Comments

feat: transform orchestrator into conversational AI coordinator#84

Open
sujayjayjay wants to merge 1 commit intomainfrom
feat/78
Open

feat: transform orchestrator into conversational AI coordinator#84
sujayjayjay wants to merge 1 commit intomainfrom
feat/78

Conversation

@sujayjayjay
Copy link
Contributor

Summary

  • Transforms the orchestrator from a documentation-focused prompt to a behavior-focused intelligent coordinator
  • Adds issue creation from descriptions: user says "fix the config bug" → orchestrator creates issue + spawns agent
  • Adds agent monitoring: orchestrator can peek at tmux sessions and summarize progress
  • Adds dependency coordination: "spawn issue 8 after 5,6,7 merge" → orchestrator watches and acts
  • Adds proactive behaviors: report progress, suggest cleanup, coordinate automatically

The key insight from #78 is that the orchestrator already runs Claude Code with full CLI access - the issue was that the prompt didn't teach it to act autonomously. This PR gives the orchestrator clear behavioral instructions for:

  1. Creating properly-formatted issues from natural language descriptions
  2. Monitoring agents by peeking at their terminal output via tmux capture-pane
  3. Coordinating dependencies by checking PR states and spawning when ready
  4. Being proactive about status updates and cleanup suggestions

Closes #78

Test plan

  • Build passes (pnpm build)
  • Lint passes (pnpm lint)
  • All tests pass (pnpm test)
  • Manual test: Run ao start and verify the generated CLAUDE.orchestrator.md has the new conversational format
  • Manual test: Verify orchestrator can create issues from descriptions and spawn agents

🤖 Generated with Claude Code

Transform the orchestrator from documentation-focused to behavior-focused:

- Issue creation from descriptions: "fix the config bug" creates issue + spawns agent
- Agent monitoring: peek at tmux sessions and summarize progress
- Dependency coordination: "spawn 8 after 5,6,7 merge" watches and acts
- Proactive behaviors: report progress, suggest cleanup, coordinate automatically

The orchestrator now acts as an intelligent assistant that handles all the
ao spawn / gh issue create plumbing so users can describe work conversationally.

Closes #78

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Copy link

@cursor cursor bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.

gh pr list --repo ${project.repo} --state open
gh pr view <number> --repo ${project.repo} --json state,reviews,statusCheckRollup
gh pr merge <number> --repo ${project.repo} --squash
\`\`\``);
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Command reference shows GitHub issue commands unconditionally

Medium Severity

The "Creating issues from descriptions" section correctly checks isGitHub and provides either gh issue commands or Linear-specific instructions. However, the "Command reference" section unconditionally includes gh issue create, gh issue list, and gh issue view commands regardless of the tracker type. For Linear projects, the orchestrator receives contradictory instructions — one section says "Use the Linear CLI" while the reference section lists gh issue commands.

Additional Locations (1)

Fix in Cursor Fix in Web

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.

feat: orchestrator session should be an AI agent, not a dumb shell

1 participant