agentleFS
inxupsinxups/route-recovery/style-creator/SKILL.md

style-creator

Skill0 starsChanged 24 days agoMITView on GitHub

Create or edit STYLE rule packages with the STYLE CLI. Use when the user requests a STYLE.yaml file, a .style.md package, or conversion of coding conventions into STYLE rules. Does not apply to CSS styling or general agent skill creation.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. leetaewon/claude-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 9 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Create local STYLE packages that express the user's requested rules.

    line 8
  2. Distinctive:

    Read existing files before editing them.

    line 16
  3. Distinctive:

    Use the official STYLE documentation for specifications and examples.

    line 20
  4. Distinctive:

    Read overview and authoring for package formats and authoring workflows.

    line 22
  5. Distinctive:

    Read package, requires, budget, and

    line 23
  6. Distinctive:

    Read rules, target, action, when, and

    line 24
  7. Distinctive:

    Read native hooks when the package must run commands or scripts.

    line 25
  8. Distinctive:

    Read CLI and diagnostics for CLI usage and reported errors.

    line 26
  9. Very specific:

    Use inject to insert content, warn to issue a warning, and block to prevent execution.

    line 44
  10. Very specific:

    Verify native hook scripts and paths.

    line 52
  11. Distinctive:

    Run these commands with the package path and confirmed harnesses.

    line 56

≈ “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

STYLE creator

Create local STYLE packages that express the user's requested rules. Validate the files and preview their behavior for the confirmed target harnesses.

Confirm the requirements

Before creating or editing a package, ask the user to confirm its name, destination path, target harnesses, package format, and rule behavior, including triggers and actions. Present known values together and ask for any missing values. Wait for a reply before writing files. Do not replace confirmation with assumptions or defaults. Ask again only if these choices change.

Recommend a directory package with STYLE.yaml and the content files it needs. Offer NAME.style.md for a single independent ordinary rule. A single-file package cannot contain multiple rules or native_hooks.

Read existing files before editing them. Preserve the confirmed structure and existing content. Deliver local files unless the user requests installation, linking, or synchronization. Creating or validating a package does not authorize those operations.

Read the documentation

Use the official STYLE documentation for specifications and examples. Fetch the Markdown pages below, not the HTML site, which renders client-side. llms.txt lists every page. Read the pages needed for the task. Do not maintain a local copy of the manual.

Check command -v style, style --version, and style --help to identify the executable. Use style help COMMAND to check its options. If the binary differs from the documentation, explain the difference and use the features that version supports. If the website is unavailable, identify the details you could not verify. If the CLI is unavailable, write the confirmed package and report that validation was not run. Do not install dependencies without authorization.

Write the package

For a new directory package, run style init NAME in the intended parent directory. Use style init NAME --single for a single-file package. NAME is a package name, not a path. Edit an existing package without initializing it again. Replace generated examples with the requested content and remove unused files generated during this task.

Choose the rule target that matches the confirmed behavior.

IntentConfiguration
Persistent coding or communication conventionsUse target: system.
Reminders before or after a tool callUse target: tool and set at: before or after.
Rules for an MCP tool or an existing skillUse target: mcp with mcp: server/tool, or target: skill with skill: name.
A trigger on a normalized eventUse target: event with an event such as tool.pre.
Command or script executionUse top-level native_hooks with a mapping for each target harness.

Use inject to insert content, warn to issue a warning, and block to prevent execution. Match the requested level of enforcement. Use mode: replace only when the user intends to replace existing system instructions.

Define conditions in when using fields available for the selected tool and event. Give blocking rules specific conditions. Prepare a matching example and a non-matching example. If a compiled index matches the current package, use style explain --index ... with test payloads to inspect the result. It does not read source packages. Do not validate a new package against an unrelated index, install it without authorization, or execute commands meant to be blocked.

Set repeat: every for every execution, or use turn or session for the requested frequency. Include reason when an inject rule uses every or turn. repeat has no effect on blocking rules.

Use inline text for short content and file: prompts/NAME.md for longer content. To keep rule metadata with its body, place both in rules/NAME.md and reference the file with from. Preserve existing IDs and keep IDs unique across rules and native hooks. Declare settings in settings and reference them with {{settings.NAME}}. Supply reasonable defaults where possible and keep credentials out of the package.

Verify native hook scripts and paths. The harness receives command unchanged, without STYLE template expansion. Do not assume the package directory is the working directory. Explain any path configuration needed on other machines.

Validate and deliver

Run these commands with the package path and confirmed harnesses.

style validate ./my-style --json
style plan ./my-style --harness codex,claude-code --json

Use the .style.md path for a single-file package. Explicit harness selection permits offline planning. It does not prove the harness is installed or its installation route is available. For a requested project preview, place the global option before the subcommand, as in style --project /path/to/project plan ./my-style --harness codex --json. Use boolean flags such as --json without a value.

Check the exit code, status, diagnostics, rule mappings, and token budgets. Review degraded, partial, skipped, never-trigger, and install-error results for changes to the requested behavior. Fix file, path, and condition errors. Report capability limits without changing block to warn, increasing budgets, or setting fallback: skip to hide the problem. A native hook without a harness mapping is skipped even with fallback: error.

Report the file paths, rule behavior, validation results, and relevant harness differences. Identify checks you did not run. A preview does not install or activate the package. The --json flag controls output format and does not make a command a dry run.

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.