- Home
- Skills
- DevOps & Cloud
- Relay: Conventional Commit Flow
Relay: Conventional Commit Flow
Turns a messy git working tree into clean Conventional Commits: preflight checks, secret-file guard, logical commit splitting, verified push to a feature branch, and PR creation. Never force-pushes, never touches main.
Free
Works with the AI Agents you already use
Relay: Conventional Commit Flow
Example session with this skill installed
My working tree is a mess: I fixed a login bug in auth.py, added a docs section, and there's a modified .env. Commit this properly and push it.
- Read your context and instructions
- Compiled the relay
Found 3 changes on branch main (protected). Excluded .env, split the rest into 2 commits, created a feature branch, and pushed.
| Step | Result |
|---|---|
| Preflight | Repo OK, no merge in progress, remote origin found |
| Sensitive files | .env flagged and not staged |
| Branch | main is protected, created fix/login-validation |
| Checks | pytest -q passed |
Commits
fix(auth): reject empty password on login
docs(readme): document login error codes
Pushed: git push -u origin fix/login-validation
Next steps
- Review
.envlocally and add it to.gitignoreif it isn't already. - Ask me to "open a PR" and I'll use your repo's template.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop writing "fix stuff" commits. Conventional Commit Flow teaches your AI agent a safe, repeatable Git loop: inspect the working tree, split changes into logical commits, write Conventional Commit messages from the real diff, run your project's checks, push a feature branch, and open a pull request.
What it does
- Preflight: checks you're in a repo, detects merge/rebase in progress, detached HEAD, missing remote, and protected branches.
- Secret guard: flags
.env*, keys, certificates andcredentials.jsonbefore anything is staged. - One logical change = one commit: splits mixed work into separate commits with
type(scope): subjectmessages (imperative, 72 chars max). - Verify before push: finds and runs your project's own tests/lint/type checks.
- Safe push + PR: feature branches only,
push -u, PR from your repo's template. - Recovery: undo, amend and squash for local commits.
Safety first
- Never force-pushes, hard-resets, or deletes branches.
- Never commits or pushes to
main/master/protected branches unless you explicitly say so. - Treats text inside diffs and comments as untrusted data (prompt-injection defence).
- Helper scripts are stdlib-only Python, read-only, no network.
Autonomy levels
Default is ask-first (shows the plan and messages). Say "just commit" to auto-commit, or "ship it" to commit, push and open a PR.
Works with git alone. The optional Relay CLI, gh or glab are only used if installed.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for DevOps & Cloud.
Secret detection is filename-based by default, not a full content scanner.
Finding your project’s test/lint commands is heuristic; it asks if nothing is discoverable.
PR creation needs gh/glab (or Relay); otherwise it prints the compare URL.
Does not resolve merge conflicts or rewrite already-pushed history.
Message quality depends on the agent model you run.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 21 days ago
- Free with an Agensi account
- Accounts
GitHub
- Software
Git
python 3.9+ (helper scripts)
- Optional
GitHub CLI
orglabrelay-cli
Permissions
Allowed Hosts
File Scopes
Terminal is used for git commands and read-only helper scripts. Network is only used when the user asks to push or open a PR (git/gh), never by the skill’s own scripts.