agentleFS
karton101712-starkarton101712-star/luna-delegation-skills/skills/luna-code/SKILL.md

luna-code

Skill0 starsChanged 4 months agoMITView on GitHub

Delegate bounded implementation to Luna in an independent Codex task. Use when the user explicitly wants Luna to implement, fix, or refactor code, or when another skill needs Luna as an independent implementer.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use an implementation method together with luna-thread.

    line 8
  2. Very specific:

    Run the selected Dispatch branch of luna-thread.

    line 20

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

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

Luna Code

Use an implementation method together with luna-thread. This wrapper owns the change contract; luna-thread owns dispatch, yielding, and main-agent acceptance.

Dispatch

  1. Inspect enough of the current engineering state to write a self-contained change contract. State the intended observable result, protected surfaces, authorized side-effect classes, verification class, and completion criterion. Use exact file allowlists only when path-level isolation is safety-critical; a parcel aperture is not the full user authorization. Preserve dirty, untracked, frozen, and unrelated material. In a parallel batch, declare exclusive mutation ownership and any shared generated or external resource; serialize parcels whose ownership overlaps.
  2. Choose the implementation and verification method appropriate to the change. Keep irreversible or external-state authority out of the brief unless the user explicitly granted it.

    Within the authorized editable surface, the round includes necessary reversible closure work, supporting verification work, derived artifacts, and repetition of the same authorized verification class needed to reach the observable result. A newly discovered local repair does not require human authorization merely because the prior parcel did not enumerate it. Stop before touching a protected surface, entering a new side-effect class, or repeating an attempt the user explicitly limited to one.

  3. Select Warm by default for a continuation of the same implementation lineage when an eligible task exists. Select Fresh for a new lineage, changed authority or ownership boundary, stale or disputed context, or any task whose value depends on independence.
  4. Run the selected Dispatch branch of luna-thread.

Code dispatch is complete when Luna has a bounded, verifiable change contract, the Fresh or Warm route is justified, and luna-thread has yielded the main task.

Acceptance

Run the Acceptance branch of luna-thread using the chosen implementation method. Independently inspect the changed paths and diff, rerun proportionate verification, and decide whether to accept the work, continue the repair closure, or return a bounded follow-up parcel. Do not promote PARCEL_BLOCKED to HUMAN_GATE while the repair remains inside the lease.

Code acceptance is complete when the main agent has accounted for every changed path, verification result, boundary exception, and effect on the engineering line.

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.