agentleFS
AIimaginedAIimagined/Godmode/skills/godmode-replicate/SKILL.md

godmode-replicate

Skill6 starsChanged 3 months agoApache-2.0View on GitHub

Rebuild a mechanism already studied in a reference implementation, in stdlib under Godmode's own names, then pin it with a reproducing test proven able to fail, cite the reference receipt and the replicating test on the host row, and scrub private terms before committing. Use when a host or platform row is marked unverifiable while its reference sits unread or unreplicated. Not for a first survey of an unread codebase, and not for an interactive session proof.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Rebuild under Godmode's names: Write the mechanism in stdlib, in this project's own modules and vocabulary - the reference's input shape …

    line 33
  2. Distinctive:

    Read PURPOSE.md for the archive records this skill exists to close.

    line 41
  3. Distinctive:

    Never rebuild from an unreceipted or surface-only reading.

    line 45
  4. Very specific:

    Never count a pinning test that has not failed on a planted break.

    line 46
  5. Very specific:

    Never copy the reference's text, identifiers or comments.

    line 47
  6. Distinctive:

    Never name the reference in shipped text.

    line 48
  7. Very specific:

    Never mark a row verified while hooks status --matrix still refuses it.

    line 49
  8. Distinctive:

    If an assertion cannot be proved, report the unmet assertion and the next safe action rather than the outcome it was meant to prove.

    line 56

≈ “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.

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

Godmode Replicate

Outcome

A host or platform row is earned by replication, not by a note: the reference's implementing files are receipted, the mechanism is rebuilt in stdlib under Godmode's names, a reproducing test is shown to fail when the rebuild is broken and pass when it is whole, the row cites both, and the commit carries no private term.

Use

  • A host or OS row reads unverifiable while a reference implementation of that path is on record but unread.
  • A studied mechanism has been judged worth borrowing and now has to exist in this project's own code.
  • The host-capability matrix refuses because a hook host cites no reference and no replicating test.
  • A replication is written and needs its pinning test proven able to fail before it counts.

Do Not Use

  • Surveying an outside codebase that has not been read yet - that is godmode-research.
  • Confirming a host's wiring with a live protected command inside that host's own IDE - that is godmode-host-sync (hooks probe).
  • Copying a reference's code verbatim into this repository - never; the rebuild is written under Godmode's names.

Deterministic Execution Flow

  1. Confirm the reference was read, not skimmed. godmode parity --sources - the reference must show surface_only: false. For each implementing file the rebuild follows, a receipt must exist: godmode read --source <name> --path <implementing file> --root <local checkout> --lines <a-b>.
    • Fallback: a surface-only or missing reference routes to godmode-research first; nothing is rebuilt from a README.
  2. Rebuild under Godmode's names. Write the mechanism in stdlib, in this project's own modules and vocabulary - the reference's input shape and expected output, not its text.
  3. Prove the pinning test can fail. godmode plant <name> --command "<the reproducing test>" --file <the rebuilt module> --replace "<a load-bearing line>" --with "<a broken line>" - the guard must fail on the planted break, and the file is restored.
    • Fallback: a test that stays green on the break does not pin the rebuild; strengthen it and plant again before step 4.
  4. Attest it green on the real rebuild. godmode verify <name> --command "<the reproducing test>" - runs the check and attests its exit code.
  5. Cite both on the row. godmode claim "<host or path> replicated from its reference" --grade verified --cite seq:<the receipt's own sequence> --cite cmd:"<the reproducing test>" --verify (a receipt: cite is for absorb decisions; a claim cites the receipt by its sequence, or it is downgraded), then check the matrix: godmode hooks status --matrix - read-only without --write; it refuses (exit 2) while any hook host cites no reference and no replicating test.
  6. Scrub before committing. godmode privacy --repo - tracked-tree scan for home paths, emails, addresses and secret shapes, each named by path and line; and the project's deny-name check (quality/checks/no_external_source_names.py) must report no new finding for the added lines.
    • Fallback: outside a git repository privacy --repo refuses (it scans tracked files only); commit from the repository the replication lives in, never skip the scan.

Read PURPOSE.md for the archive records this skill exists to close.

Must Not

  • Never rebuild from an unreceipted or surface-only reading; step 1 comes first.
  • Never count a pinning test that has not failed on a planted break; plant comes before verify.
  • Never copy the reference's text, identifiers or comments; rewrite under Godmode's names.
  • Never name the reference in shipped text; the name lives only in the private record and the receipt.
  • Never mark a row verified while hooks status --matrix still refuses it; instead fix what the matrix names and rerun it until the row passes.

Acceptance

  • The replicating test fails on a planted break and passes on the rebuild, both attested.
  • The row cites a source receipt and the replicating test, and the privacy scan of the commit is clean.

If an assertion cannot be proved, report the unmet assertion and the next safe action rather than the outcome it was meant to prove.

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.