How this skill is triggered — by the user, by Claude, or both
Slash command
/meridian:commitThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
<HARD-GATE>
Clean commits. No surprises, no slop, no AI branding.
Run git status and git diff --staged. Understand what's about to be committed.
Check for files that shouldn't be committed:
.env, credentials, secrets, API keys.idea/, .vscode/, .DS_Store, Thumbs.db)If any staged files are ambiguous — could be intentional or accidental — use AskUserQuestion to ask whether each ambiguous file (or group) should be included. Options: "Include" / "Exclude". List the specific files in the question text. Don't assume.
If files appear to be local-only (personal config overrides, local dev scripts, iterative work):
git reset HEAD <file>git log --oneline -10 for the project's commit styleShow the files and draft commit message in plain text, then use AskUserQuestion with options: "Commit" / "Edit message" / "Cancel". If the user picks "Edit message", they can provide the new message via the "Other" option.
Execute the commit with the approved message.
If the user requested "push", "commit and push", or similar compound action:
git push to the current tracking branchAskUserQuestionAskUserQuestion. Options: "Pull and retry" / "Force push" (with warning) / "Cancel push". Do not force-push without explicit user approval.If the user only said "commit" — stop after committing. Don't ask about pushing.
When subagents commit during execute, their prompts must include:
execute, debug, respond, or direct invocationGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub kodingdev/meridian