Le skill french-language impose les accents, la grammaire et la typographie française sur tous les fichiers générés : markdown, SVG, Mermaid, PlantUML, CSV, JSON, et plus.
The local-branches-status skill gives AI assistants a structured process for reporting every local branch with sync state, main diff, worktree path, last activity, and a content description.
The verify-readme-features skill systematically checks that every feature claim in your documentation is backed by actual implementation in the codebase.
LLM coding assistants are fast, but they tend to over-engineer, refactor things nobody asked for, and hide assumptions. The karpathy-guidelines skill — based on Andrej Karpathy’s observations — installs four behavioral guardrails that make sessions noticeably more disciplined.
The pin-github-actions skill automates SHA pinning of GitHub Actions, handling API lookups, tag dereferencing, and Dependabot setup.
The conventional-commit skill teaches AI coding assistants to write structured, meaningful commit messages following the Conventional Commits spec.
The diataxis skill for agent-skills brings the Diataxis documentation framework into AI coding sessions, keeping docs well-structured.
