gh-implement-plan
Skill5 starsChanged 51 days agoMITView on GitHub
Implement a GitHub plan issue step by step with checkpoints. Use when the user asks to "implement the plan", "start working on the plan", or "execute the plan".
In this repository7 other files it gives its agents.
- gh-close-planSkill
- gh-create-planSkill
- gh-create-prSkill
- gh-read-planSkill
- gh-update-planSkill
- handoverSkill
- prompt-sessionSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Prepare branch: If on main/master, fetch the latest and create a feature branch with a descriptive name derived from the plan issue title.
- Distinctive: line 21
Never commit directly to main/master -- always work on a feature branch.
- Distinctive: line 22
Never push to remote -- only stage and commit locally.
- Distinctive: line 23
If a step is unclear or seems wrong, stop and ask the user for clarification instead of guessing.
- Distinctive: line 24
If a step fails or produces unexpected results, stop and explain what happened before continuing.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Implement the plan from GitHub issue $ARGUMENTS (issue URL or owner/repo#number). If no argument is given, use the issue referenced earlier in this conversation. If no issue can be determined, ask the user.
- Prepare branch: If on main/master, fetch the latest and create a feature branch with a descriptive name derived from the plan issue title. If already on a feature branch, continue on it.
- Read the issue and identify all steps (checkboxes). Steps may be in a separate Steps comment (new format) or in the issue body (old format). Check both. Before starting implementation, think about the design: What abstractions are needed? Where should the boundaries be? Is there a simpler approach than what's described? Flag any design concerns to the user before coding.
- Present the steps to the user and confirm which step to start with. If a durable-memory CLI is configured, sync it here so recall in step 4 works off a fresh mirror.
- For each step, in order:
a. If a durable-memory CLI is configured, query it with the step's distinctive terms before implementing. See the memory tool's own integration doc for mechanics.
b. Implement the change.
c. Stage and commit the changes with a descriptive commit message. Do not reference the issue. Write the message for a reader who has only this repo and was not in this session: no session-local shorthand, and no references to the plan, other repos, other copies of a file, or tooling that does not exist in this repo. Reread it as that reader before committing.
d. Check off the step (
- [x]) on the issue. Update wherever the steps are found (Steps comment or body). - After all steps are complete, present a summary of what was done for each step (files changed, key decisions). Tell the user to review the commits and push when ready.
Rules:
- Never commit directly to main/master -- always work on a feature branch.
- Never push to remote -- only stage and commit locally.
- If a step is unclear or seems wrong, stop and ask the user for clarification instead of guessing.
- If a step fails or produces unexpected results, stop and explain what happened before continuing.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.