agentsclimarketplace
Sign in

Klaussy init

Skill steph-dove/klaussy-agents/skills/klaussy-init

Use when the user wants to make this repository AI-agent-ready. Scaffolds per-agent conventions (CLAUDE.md / GEMINI.md / AGENTS.md / Cursor rules / Copilot instructions, path-scoped via klaussy-repo-conventions), repo-namespaced skills, settings, hooks, and a PR template for Claude Code, Gemini CLI, Cursor, Codex, GitHub Copilot, Google Antigravity, Cline, and Aider — and runs once per repo.From its SKILL.md

Install
npx -y skills add steph-dove/klaussy-agents --skill klaussy-init

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.2 KB, 970 tokens by cl100k_base, as published. Nobody here has run it

Klaussy init

Scaffold AI coding-agent boilerplate for the user's project by running klaussy. By default this covers all eight supported agents — Claude Code, Gemini CLI, Cursor, Codex, GitHub Copilot, Google Antigravity, Cline, and Aider — from a single shared set of conventions.

Steps

  1. Confirm klaussy is available. Run klaussy --version. If the command isn't found, install with pipx install klaussy (preferred) or pip install --user klaussy. Re-verify with klaussy --version afterward.

  2. Detect the base branch. Klaussy will prompt interactively if not given, which blocks the agent. Pre-detect via git branch --list dev develop main master | head -1 and pass --base-branch <detected>. If none of those exist, fall back to whatever git symbolic-ref refs/remotes/origin/HEAD returns, or main.

  3. Run klaussy init. Pass through $ARGUMENTS if the user supplied any (e.g. --force, --skip-enrich):

    klaussy init --base-branch <detected> $ARGUMENTS

    If .claude/skills/ or ./CLAUDE.md already exists, ask the user whether to pass --force rather than assuming. The migration in klaussy init removes legacy .claude/commands/*.md files only if it owns them (via the .klaussy-version marker) — user-authored commands are left alone.

    Agent targets. By default klaussy scaffolds all eight supported agents — Claude Code, Gemini CLI, Cursor, Codex, GitHub Copilot, Google Antigravity, Cline, and Aider — from the same conventions. Each gets its native conventions file (e.g. CLAUDE.md, GEMINI.md, AGENTS.md for Codex/Antigravity, .github/copilot-instructions.md, .cursor/rules/, .clinerules/conventions.md, CONVENTIONS.md for Aider), plus the bundled skills in its native skills directory and permissions — except Aider, which is conventions-only (it has no skills or hooks mechanism). To narrow to specific agents, pass --agents claude,gemini (any subset). If the user only uses Claude, suggest --agents claude.

  4. Verify output. The paths below are Claude Code's; each other selected agent gets the equivalent in its own dir (.gemini/, .cursor/, .agents/ for Codex, .github/ for Copilot) plus its native conventions file. Confirm these landed:

    • ./CLAUDE.md at the repo root (with project-wide content)
    • .claude/rules/<glob-stem>.md for each path-scoped rule bucket klaussy-repo-conventions detected (zero or more files; some repos won't have any)
    • .claude/skills/<repo>-<skill>/SKILL.md for 16 skills (review, precommit, plan, debug, implement, refactor, test, fix, pr, commit, explain, humanize, new-worktree, adr-generator, security-audit, slop-coded)
    • .claude/settings.json
    • .github/PULL_REQUEST_TEMPLATE.md (only if the repo didn't already have one)
    • .gitignore updated with klaussy output exclusions

  5. Report. Tell the user which skills were created (point at <repo>-plan and <repo>-review as the high-leverage ones), summarize what's in CLAUDE.md and any .claude/rules/*.md files, and suggest reviewing them before committing.

When NOT to use

  • The repo is already klaussified at the same klaussy version — klaussy init will be a near-no-op; suggest the klaussy-update skill instead if the user actually wants to refresh.
  • The user wants a different scaffold tool (cookiecutter, an org-internal generator) — klaussy targets Claude Code, Gemini CLI, Cursor, Codex, Copilot, Antigravity, Cline, and Aider, and may not fit other setups.
  • The repo is empty or has no committed code yet — klaussy infers conventions from the existing code; an empty repo gets generic boilerplate that the user may want to defer until there's something to detect from.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most project setup skills give in 970 tokens

Counted across 1,553 of the 3,091 authors here whose files we hold, read 2026-09-06

  • Write the configuration filein 36 of 1553
  • Create the directory structurein 35 of 1553, across 33 files
  • Verify the setupin 31 of 1553, across 28 files
  • Run the setup scriptin 30 of 1553, across 29 files
  • Pre-determine the required sample sizein 29 of 1553, across 12 files
  • Check if the configuration already existsin 29 of 1553
  • Document every testin 26 of 1553, across 10 files
  • Start with a hypothesisin 26 of 1553, across 11 files
  • Ask one question at a timein 22 of 1553
  • Test a single variable per testin 21 of 1553, across 9 files
  • Read product marketing context before asking questionsin 19 of 1553, across 8 files
  • Do not peek and stop earlyin 18 of 1553, across 7 files

Said here and by no other author read

  • Run klaussy --version to confirm availability
  • Install klaussy with pipx if missing
  • Run klaussy init with detected base branch
  • Pass through any user supplied arguments
  • Confirm output files have landed successfully

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.