.claude/skills/ralph-registry-setup/SKILL.md
Inspect the repository and interactively create or update ralph.toml so Ralph has canonical test, build, lint, and format commands without depending on a package manager
npx skillsauth add peintangos/deep-agents-example ralph-registry-setupInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Arguments: $ARGUMENTS
Create or update ralph.toml as the canonical Ralph command registry for the current repository.
The registry should map Ralph command roles to the repository's existing commands. Keep the registry stack-agnostic: Ralph standardizes the role names, while each repository keeps its own task runner.
Use these keys:
test_primarytest_integrationbuild_checklint_checkformat_fixUse the literal string "N/A" when a role does not apply.
Read these first:
ralph.toml if it already existsCLAUDE.mddocs/architecture.mdThen inspect common task-runner files only as needed, such as:
package.jsonMakefilejustfilepyproject.tomlCargo.tomlgo.modcomposer.jsonFind the best existing command for each Ralph role.
Selection rules:
ralph.toml if it already contains a valid mappingAsk the user only when a role is missing or multiple commands are plausible.
If $ARGUMENTS contains constraints such as "do not use npm as canonical", honor them by keeping the registry as the source of truth and treating npm, make, just, cargo, pytest, and similar tools as underlying implementations only.
Update ralph.toml with the final mappings.
Rules:
"N/A" explicitly instead of leaving keys blankAfter updating the registry, make sure these files point to it as the source of truth:
CLAUDE.mddocs/architecture.mdIf those docs already list concrete commands, rewrite them in terms of the registry roles and note the current underlying commands only as implementation details.
Summarize:
ralph.toml mappings"N/A"ralph.tomldata-ai
監査レポートを Zenn 技術記事向けの文体 (だ/である調 + 比較表 + 階層化された見出し) に整えるスキル。`src/reporter.ts` の骨組みを LLM が整形してレポート (`out/*.md`) を仕上げるときに読み込む。
testing
OSS リポジトリの既知脆弱性 (OSV / GHSA) を照合し、重大度と影響範囲を分類するスキル。セキュリティ監査を実行するときに読み込む。
data-ai
OSS リポジトリのメンテナンス健全性 (リリース頻度・Issue 対応速度・放置 PR) を定量的に評価するスキル。メンテナンス監査を実行するときに読み込む。
tools
OSS リポジトリのメインライセンスを特定し、商用利用制約・依存互換性・NG ライセンスの検出までを行うスキル。ライセンス監査を実行するときに読み込む。