- Home
- Skills
- Workflow Automation
- reviewed git publish
Works with the AI Agents you already use
reviewed git publish
Review, freeze, and execute atomic Git commits with bilingual messages and SHA-256 plan verification.
Free
reviewed git publish
Example session with this skill installed
Review my current changes in the repo and propose a plan to commit them. I have some fixes in auth.py and a new README.md.
- Read your context and instructions
- Compiled the reviewed git publish
I have inspected the repository. You have 2 modified files and 1 untracked file. I have grouped these into two atomic commits and excluded the local .env file detected in your worktree.
Proposed Publication Plan
Target Branch: main
Remote: origin
Staging Mode: default
Commit 1
- Paths:
src/auth.py - Conventional Message:
fix(auth): correct token expiration logic - Chinese Explanation: 修复身份验证中的令牌过期逻辑。
- Risks: Low. Logic change only.
Commit 2
- Paths:
README.md,docs/setup.md - Conventional Message:
docs: update installation instructions and contribution guide - Chinese Explanation: 更新安装指南和贡献指南。
- Risks: None. Documentation only.
Excluded Paths
.env(Sensitive: local environment variables)build/(Generated: build artifacts)
Plan Digest: sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855
WAITING_FOR_USER
To proceed, please reply with:
COMMIT PLAN APPROVED sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855
Next steps
- Verify the file groupings and messages above.
- Ensure no sensitive files are missing from the excluded list.
- Provide the approval phrase to trigger the validation hooks and execution.
Connects securely to your tools. The creator never sees your data.
About this skill
Git workflows are often messy, leading to bloated commits, accidental secret leaks, or broken remote builds. This skill enforces a strict, review-first publication cycle that ensures every commit is atomic, verified, and explicitly authorized.
What it does
- Atomic grouping organizes changed paths into coherent, dependency-aware commit sequences.
- Strict validation identifies sensitive files, build artifacts, and unrelated changes to exclude them from publication.
- Bilingual documentation generates full Conventional Commit messages in English with corresponding Chinese explanations.
- SHA-256 plan freezing creates a tamper-proof manifest of exactly what will be committed and pushed.
- Pre-commit hook verification runs validation scripts in a temporary environment to ensure changes remain stable before they hit the branch.
How it works
- Use
inspectto scan the worktree and classify every changed, staged, or untracked path. - Review the proposed commit sequence, messages, and the list of excluded files.
- Generate a frozen plan and SHA-256 digest using the
freezecommand. - Provide the explicit approval phrase with the digest to authorize execution.
- The skill executes the plan, runs hooks, verifies the final state, and pushes to the remote.
Frameworks & tools
This skill utilizes a custom Python-based execution engine (reviewed_git_publish.py) and adheres to the Conventional Commits specification. It is designed for standard Git environments and integrates with existing repository hooks and validation scripts.
Why this beats prompting it yourself
Standard prompts often fail at precise file-level staging or accidentally include sensitive local environment files. This skill uses a formal schema-based plan and a cryptographic digest to ensure the agent cannot deviate from what you actually approved.
Use cases
- Enforcing clean commit history in high-stakes production repositories.
- Preparing complex pull requests where changes need to be split into logical steps.
- Preventing accidental publication of secrets or local configuration files.
- Maintaining bilingual commit logs for international development teams.
Known limitations
Does not support partial hunk staging, submodules, or detached HEAD states. Requires a Python environment to run the included executor scripts.
Requires Python for the execution script. No support for partial hunk staging, submodules, or detached HEAD. Requires explicit SHA-256 approval for every operation.
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 any payment.
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 forever
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Requires