agentleFS
bonez-iobonez-io/re_gent/.claude/skills/style-factory/SKILL.md

style-factory

Skill4 starsChanged 6 months agoApache-2.0View on GitHub

Read captured conversations to learn how this person actually works, then propose skills tailored to them. Use to discover which repeated workflows are worth turning into skills.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read prompts across sessions, most recent first.

    line 17

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Study the user's own captured history and propose skills that fit how they actually work — not how a generic developer works.

Gather

rgt sessions --format json
rgt log --json --session <session-id> -n 200

Read prompts across sessions, most recent first. Twenty or thirty sessions is plenty; this is about recurring shape, not completeness.

Look for

Repeated intent. The same request in different words, three or more times. Phrasing varies; intent repeats. That repetition is the signal — a task done once is a task, done repeatedly it is a workflow, and a workflow is a skill.

Standing corrections. Instructions given again and again ("use tabs", "don't add comments", "run the tests first"). Every repeat is a preference the agent keeps forgetting, and each one is a candidate skill or a project instruction.

Expensive sequences. Multi-step routines the user drives by hand each time — check the log, find the step, read the diff, run the tests. Sequences are where a skill saves the most.

Vocabulary. The words this person uses for their own domain. A good skill speaks their language, not the framework's.

Dead ends. Requests that took several turns to land. A skill that encodes the right approach up front removes the fumbling.

Propose

Three to five skills, strongest first. For each:

Name           <kebab-case>
What it does   <one sentence>
Evidence       <N sessions; quote 2 real prompts, verbatim>
Why a skill    <what it saves: repetition, sequence length, forgotten preference>
Data used      <which rgt commands it would run>

Rank by evidence, not by how clever the skill sounds. A dull skill backed by nine occurrences beats an elegant one backed by one.

Quote real prompts. The user should recognise themselves — that recognition is what makes them trust the suggestion.

Say explicitly what you did not find enough evidence for. A short honest list is worth more than five speculative skills.

Then

Offer to build any of them with the skill-factory skill. Do not write the files unless asked.

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.