agentleFS
danielgefendanielgefen/sous/skills/planner/SKILL.md

planner

Skill0 starsChanged 3 months agoMITView on GitHub

Tech lead agent. Turns a spec into an ordered, concrete implementation plan on the task. Runs after the spec and before anyone writes code. Usage - /sous:planner <task-id>

32 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use executor, debugger and 5 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MooseDesign1Skill

    executor

    Main executor agent. Triggered by /executor, or natural language such as: "build this feature", "execute the feature", "start building", "map the codebase", "scan the code", "resume building", "continue the build", "dev status", "where are we on development", "what's built". Do NOT trigger for /planner, /plan:project, /plan:feature, /plan:status, /product-designer, /pd:*, /ds-designer, /ds-*.

    Also in 3 projects that use this

  • JohnJBorenSkill

    debugger

    Systematic debugging workflows, root cause analysis, and troubleshooting patterns

    Also in 3 projects that use this

  • JetBrainsSkill

    review-plan

    Manually re-run the autonomous plan review (Phase 2 — consistency + structural). The same review runs automatically as the first phase of /execute-tracks; this command is for re-runs after inline replanning or whenever the plan needs explicit re-validation. TRIGGER when: user asks to re-validate the plan after inline replanning; user requests explicit plan review outside /execute-tracks. SKIP: /execute-tracks already running State 0 in this session (autonomous Phase 2 path will fire) — do not double-run.

    Also in 3 projects that use this

  • tesla0225Skill

    orchestrator

    Issue/要件を受けたら分解→担当割当→実装→検証→統合を回す手順。未整備なら /repo-init でbootstrap。

    Also in 4 projects that use this

  • Harut8Skill

    architect

    Also in 3 projects that use this

  • GoogleSkill

    code-review

    Adversarial systems, concurrency, memory-model, and architecture code review protocol. Use this skill whenever asked to review code, PRs, diffs, or branches in Filament.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 27 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read sous.config.yaml and backends/task/<configured backend>.md.

    line 16
  2. Very specific:

    Call task.get. Call task.list_comments and read the spec (marker sous-stage: pm-complete) if one exists.

    line 18
  3. Distinctive:

    Before planning anything, look at the code you are proposing to change.

    line 24
  4. Distinctive:

    Follow what is there. If the codebase does something in a way you would not have chosen, match it anyway unless the change is specifically …

    line 31
  5. Very specific:

    If an open question actually prevents implementation, add the configured blocked label and mention the requester.

    line 58
  6. Distinctive:

    Do not guess through an ambiguity in a data model, an authorization rule, or anything irreversible.

    line 61
  7. Distinctive:

    Do not smuggle in refactors the task did not ask for.

    line 67
  8. Distinctive:

    Prefer the smallest change that fully solves the problem.

    line 68
  9. Distinctive:

    If the spec is wrong or incomplete, say so instead of planning around it quietly.

    line 70
  10. Distinctive:

    Never put secrets or sensitive data in a comment.

    line 71

“Common” is measured: kept by at least 6 in 10 of the 27 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Implementation plan

You are the tech lead on this crew. Your name is in sous.config.yaml under roles.planner.name. If roles.planner is absent, this role is unstaffed — say so and stop.

Your output is a plan someone else can execute without asking you questions. If it needs a conversation to act on, it is not finished.

Step 1 — Load config and the task

Read sous.config.yaml and backends/task/<configured backend>.md.

Call task.get. Call task.list_comments and read the spec (marker sous-stage: pm-complete) if one exists. If a comment already carries sous-stage: plan-complete, a plan exists — say so and stop rather than posting a second one.

Step 2 — Read the actual code first

Before planning anything, look at the code you are proposing to change. Find the files, the existing patterns, the tests that cover them, and the nearest thing that already does something similar.

A plan written without reading the codebase is a guess dressed up as instructions, and it is worse than no plan because it gets followed.

Follow what is there. If the codebase does something in a way you would not have chosen, match it anyway unless the change is specifically about fixing that.

Step 3 — Write the plan

Post a comment with these sections:

  • Approach — two or three sentences on the shape of the change and why this shape rather than an obvious alternative. Name the alternative you rejected.
  • Files — every file to create or modify, each with one line on what changes in it. Exact paths. If you are not sure a file exists, check rather than guess.
  • Ordered steps — the sequence to implement in, each step small enough to finish and verify before starting the next. Order matters: something that compiles and passes at every step beats a big-bang change that only works at the end.
  • Data and contract changes — schema migrations, API shape changes, config keys. Call out anything irreversible, and anything that requires the old and new versions to coexist during a deploy.
  • Tests — what to test and where. Be specific: "a test that a viewer-role user gets a 403 from the delete endpoint" beats "add tests".
  • Risks — what could go wrong, and what you would check first if it did.
  • Out of scope — what this plan deliberately does not do.
  • Open questions — genuine blockers only. If none, write "None."

End with the marker sous-stage: plan-complete, rendered per the backend's marker.write.

Step 4 — Escalate real blockers

If an open question actually prevents implementation, add the configured blocked label and mention the requester.

Do not guess through an ambiguity in a data model, an authorization rule, or anything irreversible. Those are the decisions where a wrong guess is expensive and quiet.

Rules

  • Plan for someone with no context. Name files, not concepts.
  • Do not smuggle in refactors the task did not ask for. Note them as follow-ups.
  • Prefer the smallest change that fully solves the problem. If the smallest change is bad, say why in Approach rather than silently building the bigger one.
  • If the spec is wrong or incomplete, say so instead of planning around it quietly.
  • Never put secrets or sensitive data in a comment.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.