- Home
- Skills
- Workflow Automation
- devflow commit
Works with the AI Agents you already use
devflow commit
Turns your working tree into atomic Conventional Commits, one logical change each, with the ticket reference taken from the branch name. Shows the plan first, validates every message, never pushes.
Free
devflow commit
Example session with this skill installed
I have some changes in authStore.ts and a fix in RemoteLoader.tsx. My branch is feat/PROJ-55-auth-sync. Can you commit these?
- Read your context and instructions
- Compiled the devflow commit
I've analyzed the working tree and the branch name feat/PROJ-55-auth-sync. I found two logical changes. Here is the plan:
Commit Plan
Branch: feat/PROJ-55-auth-sync
Ticket: PROJ-55
- feat(auth): add Zustand auth store
shell/src/store/authStore.ts
Message
feat(auth): add Zustand auth store
Refs: PROJ-55
- fix(shell): prevent RemoteLoader remount on route change
shell/src/components/RemoteLoader.tsx
Message
fix(shell): prevent RemoteLoader remount on route change
Refs: PROJ-55
Next steps
- Type "approve" to execute these commits.
- Specify if you want to combine these into a single commit instead.
- Tell me if any files should be excluded from this round.
Connects securely to your tools. The creator never sees your data.
About this skill
Struggling with messy working trees and illegible git histories? Stop dumping unrelated changes into a single "wip" commit. This skill enforces atomic commit discipline and Conventional Commits standards automatically.
What it does
- Atomic grouping identifies logical changes to group features, tests, and docs together while separating unrelated fixes.
- Ticket extraction parses your branch name to automatically append the correct Jira or GitHub ticket reference to every commit footer.
- Message validation enforces the Conventional Commits 1.0.0 specification including type, scope, and character limits.
- Interactive planning presents a full commit strategy for approval before staging a single file.
- Safe staging prevents accidental inclusion of secrets,
.envfiles, or local credentials.
How it works
- Analyze state by checking the current branch, status, and diffs to identify pending changes.
- Determine conventions by reading project-specific
commitlintconfigs or recent git history to match existing styles. - Propose plan showing exactly which files go into which commit, along with the generated message and ticket footer.
- Execute and validate each commit sequentially, running internal checks to ensure messages pass linting before finalizing.
Frameworks & tools
This skill requires git and bash. It integrates with commitlint configurations and recognizes standard ticket patterns in branch names.
Why this beats prompting it yourself
Prompting an LLM to "write a commit message" often results in flowery or inaccurate descriptions. This skill uses dedicated bash scripts to strictly validate message length, format, and ticket references, ensuring your history is machine-readable and professional without manually copy-pasting diffs.
Use cases
- Splitting a large "end of day" working tree into four clean, logical commits.
- Formatting commit messages to trigger automated changelog generators.
- Ensuring every commit in a PR is linked to the correct Jira ticket automatically.
- Cleaning up a branch's history before a final review.
Known limitations
Does not support interactive git modes like git add -p. Requires a branch name that follows recognized ticket patterns to auto-link issues.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.
Cannot use interactive git commands like add -p. Will not push to remote. Requires git and bash installed. Cannot guess tickets if not in branch name.
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 1 day ago
- Free with an Agensi account
- Software
Git