agentleFS
TsukiNi22TsukiNi22/ai-utils/skills/routers/dev/SKILL.md

dev

Skill1 starsChanged 3 days agoMITView on GitHub

Global development router - invoked manually with /dev, it inspects the request, the project and the languages involved, then loads the matching skills of the user (using the tables of the cpp, python, git, doc, style, audit and legal routers: cpp-project, cpp-class, python-project, python-class, python-style, python-comments, ci-cd, pre-commit, rtk, interactive-questions, clarify, coverage, benchmark, cpp-style, coding-style, cpp-comments, comments, cmake-style, cpp-tests, tests, audit-bugs, audit-quality, libutils, libutils-exception, libutils-install, libutils-setup, git-conventions, readme-style, wiki-style, html-doc, html-style, explain-doc, report, license, audit-deps) before doing the task.

55 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use metrics-graphana, control-flow and 10 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.
  • slopusSkill

    metrics-graphana

    Query and manage Grafana dashboards and Prometheus metrics for Happy infrastructure. Covers grafanactl CLI usage, direct Prometheus queries through Grafana proxy, and dashboard-as-code workflows. Use when user asks about metrics, dashboards, monitoring, Grafana, Prometheus, or wants to add/modify panels.

    Also in 3 projects that use this

  • slopusSkill

    control-flow

    Analyze and design control flows and data structures. Produces compact ASCII tree diagrams showing triggers, call chains, payload shapes, state mutations, and re-render effects. Use when user asks to diagram, trace, visualize, or design a flow or data structure.

    Also in 3 projects that use this

  • MicrosoftSkill

    sessions

    Core principles and workflow router for changes to the Agents Window under src/vs/sessions.

    Also in 3 projects that use this

  • AnthropicSkill

    office-hours

    Set up and run Claude Code office hours, including the calendar event when the champion's own calendar connector allows it. Use when a champion wants a regular time for teammates to bring problems.

    Also in 3 projects that use this

  • nicepkgSkill

    team

    根据任务快速组建临时 AI Agent 团队协作。自动从 .claude/agents/ 中选择最合适的成员组队。

    Also in 3 projects that use this

  • Microsoft AzureSkill

    test

    Also in 5 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 37 edited versions we read keep these.

  • 1

    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:

    Every audit, review, analysis or summary delivered to the user (including the "no request" state of the project) goes through report …

    line 94
  2. Distinctive:

    Follow the loaded skills strictly.

    line 97
  3. Very specific:

    Never commit/push/tag/release/open a PR without an explicit request (git-conventions).

    line 98
  4. Very specific:

    Start the answer with one line: Skills : <list> (<reason>), then do the task.

    line 99

“Common” is measured: kept by at least 6 in 10 of the 37 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

Development router

<name> (a skill) = the folder of that skill: ~/.claude/skills/<name> (setup.sh) or ${CLAUDE_SKILL_DIR}/../../*/<name> (plugin of the marketplace).

Request given with /dev: $ARGUMENTS

Single entry point: the user only invokes /dev (with or without a request) and Claude decides which skills to use. Do not answer from this file: load the skills with the Skill tool first, then work.

1. Understand the request

The text given with /dev, otherwise the latest request of the user. With nothing at all: inspect the project (step 2) and propose the next useful actions (uncommitted changes to commit, missing tests/docs, outdated README/CHANGELOG, new libutils version...), without changing anything.

2. Inspect the context (in parallel, quick)

  • git status --short, git log --oneline -5, git remote -v.
  • Languages / build: CMakeLists.txt, *.cpp/*.hpp (C++), package.json, pyproject.toml, Cargo.toml...
  • libutils: find_package(utils in the CMake or utils:: in the sources.
  • Docs: README.md, CHANGELOG.md, docs/.

3. Route

The category routers (cpp, python, setup-project, git, doc, style, legal, audit) are manual-only: don't call them with the Skill tool. Read their decision table when the case is not obvious (<router>/SKILL.md), then load the specific skills with the Skill tool:

NeedLoad (Skill tool)Detailed table
new C++ projectcpp-projectcpp
C++ code, classes, main, architecturecpp-class, cpp-style, cpp-commentscpp
CMake, tests CMake, packagingcmake-stylecpp
libutils APIs / exceptionslibutils, libutils-exceptioncpp
install / update libutils on the computer, add it to the projectlibutils-install, libutils-setupcpp
unit tests (C++: GoogleTest like libutils)cpp-tests (C++) or tests (any language, asks the framework)cpp
test coverage: measure, what is not tested, fill itcoverage (C++, cpp-tests projects)audit
performance: benchmark, profiling, what is slow, optimizebenchmarkaudit
CI / CD, GitHub workflows, actions, Docker images of the CIci-cdgit
pre-commit hook (xstyle)pre-commitgit
compact a text / output for an AI, save tokens, add an --rtk mode to a programrtk-
choices / ideas / questions to ask the user ("propose en qcm", "demande-moi", several approaches)interactive-questions-
request unclear / ambiguous / incomplete, "clarifie", "vérifie que tu as compris", ask before actingclarify (uses interactive-questions)-
ideas, recommendations, an opinion on ideas, variants ("des idées ?", "ton avis", "améliore cette idée", brainstorm)ideation (uses interactive-questions)-
hand the work over to another session / machine / model ("export", "passation", "handoff", PASSATION.md)export (uses rtk)-
set up the context before working ("setup le contexte", read the repo and the old / archived sessions, load the useful skills / MCP)setup-context (uses onboard when installed, interactive-questions)-
no interruption until the end ("ne pose aucune question", "fais tout d'une traite", "pose tes questions à la fin")ask-at-end (asks at the end with interactive-questions / ideation)-
stop cleanly now and resume later ("stop tout proprement", "mets en pause", "on reprendra plus tard", /pause)pause (checkpoint in ~/.claude/pauses)-
resume a paused task ("continue", "reprends" after a pause, a checkpoint waiting for the project)continue-work-
write in the user's own voice ("écris comme moi", "avec mes mots", a message / note / prompt in his style)write-like-me-
set up the context of a project / folder before working on it (read the repo, its git history, the earlier sessions that touched it: "onboard", "lis tout le repo", "prends connaissance du projet"; the project only, the skills / MCP of the session too: setup-context)onboard (uses rtk)-
make a presentation / slides / a deck / a PDF or PPT from a plan or an idea ("fais une présentation", "un ppt")presentation (uses interactive-questions)-
give every git / repository permission from the prompt ("tu as tous les droits", "commit/push sans demander", /full-permission)full-permission-
set up a whole new project (architecture + tests + license + git + CI/CD + README)the steps of setup-project (its table, skills loaded in that order)setup-project
several / all audits of the project at oncefollow the audit router (checklist of audits + format)audit
bugs / UB / crashes / sanitizers auditaudit-bugscpp
security audit, flaws, vulnerabilities, hacking, memory errors fixed until cleanaudit-security-
cleanliness / conventions audit with a PDF reportaudit-qualitystyle
comments of any language, file headerscomments (+ cpp-comments)doc
a deliverable as PDF (report, study, benchmark)reportdoc
license (choose, add, replace, compatibility)licenselegal
dependencies: licenses to credit / restrictive / paid, vulnerabilitiesaudit-depslegal
legal compliance check: GDPR / RGPD, cookies, mentions légales, CGU, RGAA, privacy, erasure, AI Act, licenses and mentionslegal-checklegal
make it compliant: privacy policy, legal notice, cookie banner, terms, accessibility statement, data deletion / exportlegal-setuplegal
several languages / translate a project (multilingual-project) or a website (multilingual-website + multilingual-project)multilingual-project, multilingual-website-
dark / light / system modes, themes, high contrast: a project (color-modes-project) or a website (color-modes-website + color-modes-project)color-modes-project, color-modes-website-
style only (format / review like the user)cpp-style, cpp-comments, coding-style (other languages), cmake-style, comments by file typestyle
Python codepython-style + python-commentspython
new Python class / module / tool / exception / test filepython-classpython
new Python project / package / toolpython-projectpython
code in another language than C++ (shell, Lua, JS/TS, Rust, C...)coding-style + commentsstyle
README, Markdown docsreadme-styledoc
GitHub wikiwiki-styledoc
HTML docs, project graphhtml-docdoc
documentation site with search (claude-cluster docs style)better-html-doc (+ better-explain-doc / better-protocol-doc for its explanation / protocol pages; one page in that look: better-html-style)doc
explanation page (how something works), simple + technicalexplain-docdoc
protocol of a networking project (frames, handshake, codes, commands, storage)protocol-docdoc
any other HTML page, palette / color / fonthtml-styledoc
free space: caches, Docker, old builds, Claude Code / claude-cluster leftoverscomputer-clean-
set up / scan / refresh the creative tools (audio, image, video, 3D, simulation / VFX / rendering, their MCP servers) as skillssetup-studio (then /studio routes the creative requests)-
a creative request: sound, music, voice, image, video, 3D model, simulation, renderthe studio router when it was generated (/studio), else propose setup-studio-
commit the current diff locally ("fais le commit")git-commitgit
update every doc / site / wiki / package / install script after changes or a new versiondocs-updatedoc
commit message, branch, PR, tag, release, CHANGELOGgit-conventionsgit
several needs (ex: "add the feature, document it and prepare the commit")every matching skill, in the order of the work: code -> docs -> git-
no skill matches (other language/tool)none: say it, follow the conventions of the project-

4. Work

  • Every audit, review, analysis or summary delivered to the user (including the "no request" state of the project) goes through report (format asked once: Markdown + PDF recommended / PDF / Markdown), English by default, in audit/ at the repository root unless asked; the chat only gives the verdict and the paths.
  • Follow the loaded skills strictly; they override generic habits.
  • Never commit/push/tag/release/open a PR without an explicit request (git-conventions).
  • Start the answer with one line: Skills : <list> (<reason>), then do the task.

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.