agentleFS
Magnus-GilleMagnus-Gille/claude-skills/code-friction/SKILL.md

code-friction

Skill5 starsChanged 33 days agoMITView on GitHub

Capture a small, opt-in set of concrete code-changeability observations during a coding task and hand them to Close without affecting readiness.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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:

    Keep no more than three distinct observations for the task across all worker attempts.

    line 15
  2. Distinctive:

    Use this when a coding task encounters or benefits from a specific property of the code or its environment, such as a hidden consumer that …

    line 18
  3. Distinctive:

    Do not capture: “the task was hard” without a concrete, evidenced cause.

    line 27
  4. Distinctive:

    Do not capture: a failed network or unavailable tool as a code difficulty.

    line 28
  5. Distinctive:

    Before capture, read the Close integration and persistence rules and the bundled changeability rubric.

    line 31

≈ “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

Code-friction capture

This skill records advisory evidence about how easy or difficult a coding change was to locate, understand, and verify. It is not an agent score, quality gate, telemetry stream, or authorization to write to Munin.

Capture is off by default. Begin only when the owner has enabled it for this task or the user explicitly authorizes it. Continue noticing meaningful events during the work instead of waiting until Close: record a concrete helpful or obstructive factor when it changes how the task proceeds. Keep no more than three distinct observations for the task across all worker attempts. Quick Close can flush prepared records but never invents observations; full Close can finalize an assessment.

Use this when a coding task encounters or benefits from a specific property of the code or its environment, such as a hidden consumer that changes the implementation path or a deterministic test that makes verification direct. Do not use it for ordinary Q&A, routine steps without a meaningful friction or helpful factor, generalized agent performance claims, or a task without explicit opt-in.

Examples:

  • Capture: a second known caller had to be traced before the behavior was understandable.
  • Capture: an existing deterministic test isolated the changed behavior with no additional setup.
  • Do not capture: “the task was hard” without a concrete, evidenced cause.
  • Do not capture: a failed network or unavailable tool as a code difficulty; classify the cause as environmental and leave unsupported ratings not-assessable.

Before capture, read the Close integration and persistence rules and the bundled changeability rubric. Keep summaries sanitized, short, and free of code, prompts, transcripts, private locators, raw errors, or credentials. Use opaque evidence references only. If safe minimization or evidence provenance is uncertain, omit the observation. Do not make an extra evaluator call.

Prepare evidence only with an exact writable namespace supplied by the user or owner. Persist the full Munin append arguments before sending; retry only those saved arguments and their stable idempotency key after an uncertain response. Never guess classification. Report unavailable, refused, or failed persistence as unsaved through the existing Close Persistence section. The helper is local-only; it makes no network call and changes no Close renderer or readiness result.

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.