Add coordinator profiles and role escalation ladders

[tomastaker]

Changes

Delegation now supports coordinator-specific teams. Each role has a usual model and optional strengthening steps, combining native agents with external CLIs. Profile names are user-defined; GPT, Claude and Kimi are editable examples. The current chat remains the coordinator.

• Added role profiles, explicit model and reasoning selection, native/external routing, and compatibility with existing configuration.
• Let the coordinator choose useful parallelism. Optional user limits bound concurrent workers, writers, total runs and retries; counters update atomically without estimating API costs.
• Clarified when to resume with a better brief, switch to a stronger model, isolate writers and request independent review.
• Rewrote the README with a new illustration, a single-file configuration guide, and a custom GLM team using Claude and GPT workers. The Delegate Kit name is unchanged.
• Fixed task-only resumes, legacy route argument roundtrips, and accounting for external writers when acquiring native worktree locks.
• Isolated profile tests from locally installed CLIs so they also run on clean CI hosts.

Validation

• 62 Node tests and 127 shell checks passed; GitHub CI passed on the final PR revision.
• Bash/Node syntax, ShellCheck, JSON validation, the SKILL.md size limit and git diff --check passed.
• The local README HTML preview was inspected with Playwright at 390, 768 and 1280 px.
• Independent review findings were fixed and rechecked.

External CLI integration tests use fixtures; no paid model calls were made. Supported external adapters are Codex, Claude, Gemini and OpenCode. A profile alone does not add an arbitrary CLI or a direct API integration. Existing personal skill installations need to be updated separately.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 418be050-024b-4f69-a5bc-47fd145a1021

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-08T21:01:10.101796Z	a01e014	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a01e014025

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".