docs(skills): teach CLI agent orchestration

[Waishnav]

The bundled skills previously mixed MCP instructions, TUI behavior, provider adapter details, and implementation notes. This layer rewrites them as two installable CLI guides: subagents for bounded delegation and follow-ups, and dynamic-workflows for programmable multi-agent orchestration.

Both skills focus on capabilities, use cases, structured output, polling, recovery, and concise examples. Provider-specific reference files are removed because target discovery belongs to devspace agents targets --json.

Both skill packages pass the skill validator, and the repository tests cover the revised discovery behavior.

Summary by CodeRabbit

• Documentation
  • Updated Dynamic Workflows guidance with CLI-based instructions for creating, running, inspecting, composing, and recovering resumable workflows.
  • Added guidance for JSON polling, workflow metadata, schemas, profiles, isolation, replay, and deterministic execution.
  • Expanded subagent guidance with CLI discovery, session inspection, delegation, polling, continuation, and workflow selection.
  • Removed outdated provider-specific model and effort-setting references.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 7cca0c7b-4033-46e3-9375-834356ba6a25

📥 Commits

Reviewing files that changed from the base of the PR and between 7d602ea and 411e393.

📒 Files selected for processing (9)

• skills/dynamic-workflows/SKILL.md
• skills/subagents/SKILL.md
• skills/subagents/references/claude.md
• skills/subagents/references/codex.md
• skills/subagents/references/copilot.md
• skills/subagents/references/cursor.md
• skills/subagents/references/opencode.md
• skills/subagents/references/pi.md
• src/skills.test.ts

💤 Files with no reviewable changes (7)

• skills/subagents/references/pi.md
• skills/subagents/references/claude.md
• skills/subagents/references/codex.md
• skills/subagents/references/opencode.md
• skills/subagents/references/cursor.md
• skills/subagents/references/copilot.md
• src/skills.test.ts

---

📝 Walkthrough

Walkthrough

The change replaces legacy skill guidance with DevSpace CLI instructions for Dynamic Workflows and subagents. It documents JSON execution, polling, recovery, orchestration, session continuation, and provider selection. Obsolete provider reference documentation and related test coverage were removed.

Changes

CLI Skill Guidance

Layer / File(s)	Summary
Dynamic Workflow CLI and replay guidance skills/dynamic-workflows/SKILL.md	Documents CLI execution, inspection, polling, orchestration, schemas, isolation, determinism, recovery, and recommended workflow use cases.
Subagent discovery and session guidance skills/subagents/SKILL.md, src/skills.test.ts	Documents target discovery, provider selection, task execution, session polling, continuation, and when to use Dynamic Workflows. Removes tests for the deleted provider reference.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• Waishnav/devspace#83: Added related workflow CLI and subagent orchestration features.
• Waishnav/devspace#100: Strengthened related CLI, provider, schema, and resumable-workflow contracts.
• Waishnav/devspace#118: Introduced related provider reference documentation and skill updates.

Poem

A rabbit hops through CLI lanes,
Polling workflows, checking gains.
Subagents start, then sessions wait,
Resumable paths keep tasks on slate.
Old provider notes drift away—
New skill guidance leads the way.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: updating skills documentation for CLI-based agent orchestration.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/dw-cli-skills

---

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.}

[greptile-apps]

Greptile Summary

The PR rewrites the Dynamic Workflows and subagent skills around the asynchronous JSON CLI surface and removes the former provider-specific override references.

• Adds command-oriented run, inspection, continuation, and recovery guidance.
• Consolidates target and override advice into the main subagent skill.
• Removes obsolete reference-path coverage from the skill tests.

Confidence Score: 4/5

The PR should not merge until recovery guidance makes clear that worktree-isolated calls and all later calls execute again during resume.

The rewritten skill recommends worktree isolation while omitting its non-replayable behavior, so following the documented recovery flow can unexpectedly rerun successful mutations and downstream work; the subagent polling status omission is additionally non-blocking.

Files Needing Attention: skills/dynamic-workflows/SKILL.md, skills/subagents/SKILL.md

Important Files Changed

Filename	Overview
skills/dynamic-workflows/SKILL.md	Replaces the workflow documentation with concise JSON CLI guidance, but omits the important constraint that worktree-isolated calls terminate replay reuse.
skills/subagents/SKILL.md	Documents asynchronous subagent execution and polling, but should enumerate the agent-specific terminal statuses.
src/skills.test.ts	Removes path-resolution assertions for reference files deleted by this PR; no remaining source consumer of those paths was identified.

_{Reviews (1): Last reviewed commit: "test(skill): drop provider reference ass..." | Re-trigger Greptile}

[Waishnav]

[gpt-5.4] RESPONDING ON BEHALF OF WAISHNAV

Fixed in 411e393. The workflow skill now states that a completed worktree-isolated call cannot be reused, so that call and every later call execute again during resume. The subagent skill now names starting and running as polling states and idle, error, and stopped as terminal outcomes. Both skill packages pass the skill validator.

[coderabbitai]

@Waishnav I will perform a complete review of pull request #143.

⚠️ Action not completed

Review rate limited.

---

Your included review limit is currently reached under our Fair Usage Limits Policy. This review may still proceed through usage-based billing if eligible. Your next included review will be available in 57 minutes.

[Waishnav]

@coderabbitai full review

[coderabbitai]

✅ Action performed

Full review finished.