agentleFS
5uck1ess5uck1ess/devkit/skills/executing/SKILL.md

executing

Skill4 starsChanged 4 months agoMITView on GitHub

Execute implementation plans — work through steps methodically, verify each one, keep changes small and reviewable.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 5

    Distinctive

    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
  1. Distinctive:

    Do not skip verify. Do not batch multiple steps into one change.

    line 17
  2. Distinctive:

    Don't go down rabbitholes: If you discover something unrelated that needs fixing, note it and come back later.

    line 22
  3. Distinctive:

    Verify at the boundary: After completing a step, confirm the system still works end-to-end, not just the part you touched.

    line 24
  4. Distinctive:

    Commit after each verified step.

    line 35
  5. Distinctive:

    All existing tests pass

    line 47

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Executing

The Loop

For each step in the plan:

  1. Understand — Re-read the step. What exactly needs to change?
  2. Implement — Make the smallest change that completes the step.
  3. Verify — Run tests, check behavior, confirm it works.
  4. Commit — Save the working state before moving on.

Do not skip verify. Do not batch multiple steps into one change.

Discipline

  • One step at a time. Finish step N before starting step N+1. The urge to "quickly also do" the next thing leads to half-done work.
  • Don't go down rabbitholes. If you discover something unrelated that needs fixing, note it and come back later. Stay on the current step.
  • Keep changes small. A 20-line diff is easy to review. A 200-line diff hides bugs. If a step produces a large diff, it should have been broken into smaller steps.
  • Verify at the boundary. After completing a step, confirm the system still works end-to-end, not just the part you touched.

When Things Go Wrong

  • Step is harder than expected — Stop. Re-plan. Break it into sub-steps. Don't push through with a "it'll work out" mindset.
  • Step reveals a flaw in the plan — Update the plan. Plans are living documents.
  • Tests break — Fix them now, not later. Broken tests that you "plan to fix" become permanently broken tests.
  • You're stuck for 15+ minutes — Step back. Re-read the requirements. Try a different approach. Ask for help.

Checkpointing

Commit after each verified step. Commit messages should reference the plan step:

step 3: add input validation to parseConfig

This creates a clean, reviewable history and makes it easy to bisect if something breaks later.

Done Means Done

A step is done when:

  • The implementation matches the step description
  • All existing tests pass
  • Any new behavior has tests
  • The code is clean (no TODOs, no commented-out code, no "temporary" hacks)

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.