agentleFS
kaotyprkaotypr/context-circuit/.agents/skills/cc-plan/SKILL.md

cc-plan

Skill0 starsChanged 17 days ago0BSDView on GitHub

Create Context Circuit plan folders from an approved intent or a direct request for a grounded plan.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

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

    Write a plan in the language recorded for the active member in members.yaml, and in English where none is.

    line 25

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Write grounded plan folders

Planning starts on a person's request to plan, not on approval. record approve returns planning_recommended: approval records the decision, and the next step is to recommend planning to the person and stop. A later request to plan an approved intent is the planning request, and so is one request that approves intents the person has already been shown and asks for their plans; plan each named intent then, without a stop and without asking for approval again. Asking for an intent to be written and planned in one go is not approval. In a Git workspace approval also returns commit_recommended, naming the records and ID files left uncommitted: ask the person once whether to commit and push them, or commit only; their answer is the authorization, and a refused push is handled as AGENTS.md says under delivery.

An approval answered with planning_required instead comes from a CLI older than this template. Tell the person to install or upgrade to the version the workspace pins, through .agents/skills/cc-cli/SKILL.md, and follow AGENTS.md rather than that output's text.

Write a plan in the language recorded for the active member in members.yaml, and in English where none is. A plan is quoted whole into the brief a worker receives, and that brief tells the worker that what goes into the repository is English regardless. Identifiers and the project's own domain vocabulary keep their form whatever language the prose is in, and the slug stays a lowercase ASCII slug. So do the headings: a plan's structure is English whatever its prose is, for the reason the slug is.

Compose in that language rather than translating into it — let the sentences follow that language rather than keeping English clause order, prefer the verb where English would nominalize, and keep technical vocabulary in the form engineers there actually say. Where the member records a tone in members.yaml, follow it; it overrides that guidance. .agents/skills/cc-intent/SKILL.md carries the same rules with worked examples.

Once planning is requested for an approved intent, inspect real code and create linked plan folders:

sh <root>/.context-circuit/cli record create --kind plan --slug SLUG --title TITLE --intent i001 --repo api

For a person's explicit request for a detailed plan of an already specified outcome, do the same grounding and omit --intent. Do not invent an intent. Neither planning request authorizes execution. Present the complete plan and wait for a separate request to execute it.

How many plans

One plan folder may cover one or several repositories. Use separate plans when that helps execution or delivery — work that lands in different repositories at different times, or that a person may want to run in stages.

Where plans depend on each other, record it, so the order can be derived rather than guessed:

sh <root>/.context-circuit/cli record dependencies --id p0001 --depends-on p0002

What a plan holds

plans/pNNNN-slug/plan.md holds the outcome, repository scope, approach, task order, optional dependencies, useful risks, expected checks, and a Details section. Add concern files such as api-contract.md or data-model.md only when the extra detail helps a person inspect the proposal. Link every supporting file from Details and assign it in frontmatter, relative to the plan folder:

required_files:
  shared: [data-model.md]
  by_repository:
    api: [api-contract.md]
    web: [ui-structure.md]

Omit required_files when there are no supporting files. Every regular file inside the folder besides plan.md must be assigned. Paths stay inside the folder, do not pass through symlinks, and refer to regular files. Use Markdown for text details; diagrams and assets can also live inside the folder. The entry remains a readable map of all details, including those assigned to other repositories. Present its links and material choices to the person, then revise the folder as requested. Re-present any material change made after presentation.

Explain and record newly needed files or repositories within the approved outcome. Obtain renewed approval only when the intended outcome or its success criteria materially change; a question that bears on the outcome returns to the intent record.

Revise the plan during review and present material changes again. Once execution starts, running a plan does not edit it, so a later reader can see what was intended and read the diff for what happened.

When a planner produced it

Transcribe the planner's approach, its tasks and order, and its risks and checks. Its verdict and its plan shape are answers to you, not plan content: once the records exist the shape is the records, so a plan saying it is a single plan states what its own existence already says.

A not-feasible return is a complete answer — write no plan, and report it with its reasons. A return saying the outcome or success criteria must change goes back for renewed approval. The coordinator writes every record either way.

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.