micro-loop
Skill2 starsChanged 7 days agoMITView on GitHub
Execute a simple, low-risk phased plan from `.plans/{task}/` straight through with no builds, tests, or commits between phases; verify, lint, and commit once at the end. Use when the plan is straightforward and rebuilding per phase would only waste time.
In this repository18 other files it gives its agents.
- approve-planSkill
- check-impactSkill
- context-packSkill
- design-forgeSkill
- deslopSkill
- diagnose-bugSkill
- falsifySkill
- forgeSkill
- gap-analysisSkill
- interrogateSkill
- make-fixtureSkill
- phased-loopSkill
- pick-tierSkill
- ship-planSkill
- simple-loopSkill
- taste-testSkill
- unslopSkill
- wait-whatSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 50
All phases are done and the list is clean.
- Very specific: line 52
Check: Run check and fix until it passes.
- Very specific: line 54
Test: Run test; all tests pass.
- Very specific: line 60
Sync: Run sync once (resolve conflicts, re-test, commit the merge).
≈ “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
Micro Task Loop
A fast path for simple phased plans. Trades per-phase safety checkpoints for wall-clock time: no builds, tests, or commits until every phase is done, then one verification pass and one commit. tasks.md tracks progress in real time.
Arguments: $ARGUMENTS — the plan (a .plans/<task>/ path or slug). Always run the whole plan.
Picked by pick-tier when the tier is S or M.
Read task-loop-rules.md in this skill's folder first. It holds the project commands, constraints, concepts, task-list discipline, and finishing steps all three task loops share. This file adds only what is specific to this loop.
Constraints (on top of the shared ones)
- Edit-only until the end. During phases, only edit files;
checkandtestwait for Final verification — not per-package, not per-module, not "just this once." Avoiding rebuild-thrash is the entire point of this skill. Exception: an environment or toolchain failure that blocks editing itself.
The Loop
Repeat until every task is complete. Work per phase, top to bottom.
- Check status.
Readtasks.md. Every task[x]→ go to Final verification. Otherwise start at the next task. - Gather what the current task needs: the spec section it implements and the code it touches.
Implement, one task at a time:
a. Restate the current task in one sentence.
b. Do the work — edits only (Constraint 2).
c. Tick it in
tasks.md.d. Pick the next task under the current phase header and repeat from (a). When the phase has no
- [ ]left, go to 3.5.
3.5. Audit. Re-read tasks.md. Any task in this phase not [x] means you drifted: return to step 3 with it. The list must be clean before you move to the next phase.
- Go to step 1.
Final verification
All phases are done and the list is clean. Run these in order without stopping, unless an unrecoverable error blocks progress:
- Check. Run
checkand fix until it passes. Failures can point at any phase — fix forward. - Test. Run
test; all tests pass. - Lint and format as in Finishing step 1 of the rules file.
- Commit once with a Conventional Commit message,
<type>(<scope>): <summary>(lowercase, imperative, 72 characters or fewer), only after check, tests and lints pass. If an unrecoverable failure blocks verification, stop and report instead of committing. - Sync. Run
synconce (resolve conflicts, re-test, commit the merge). - Hand back as in Finishing step 2 of the rules file.
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.

