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

godmode-evidence

Skill6 starsChanged 3 months agoApache-2.0View on GitHub

Shape a claim so its citations grade the first time: reflect it against the record, cite exit-bearing commands through `--verify`, give a hypothesis its `--refuted-by` falsifier, run falsifiers that aged past due, repair stale or loose citations, and resolve the claim once with fresh evidence. Use when a claim keeps being downgraded, a Stop gate keeps blocking on unsupported sentences, or cited evidence has drifted. Not for a second audit of a pass already accepted, and not for root-causing a failure.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 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.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

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

    line 42
  2. Distinctive:

    Never write a wrapper script to turn a state-reporting command into an exit-bearing one.

    line 46
  3. Very specific:

    Never record a hypothesis without --refuted-by.

    line 47
  4. Distinctive:

    Never re-assert a stale claim over its drifted evidence.

    line 48
  5. Distinctive:

    Never resolve a claim with only the evidence it was recorded with, or resolve one sequence twice.

    line 49
  6. Very specific:

    Never run verify --falsifiers before its --dry-run listing.

    line 50
  7. 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 57

≈ “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 Evidence

Outcome

A claim lands at the grade it asks for on the first attempt, because its citations are shaped to be graded rather than argued: exit-bearing commands run through --verify, a hypothesis names what would kill it, stale and loose citations are repaired from the record instead of re-asserted, and a claim is resolved once, with evidence it did not already have.

Use

  • A claim was downgraded (to observed or hypothesis) and the cite needs reshaping rather than a wrapper script.
  • The Stop gate keeps blocking on sentences with no claim record behind them.
  • A hypothesis-grade claim has no falsifier, or its falsifier has aged past due unrun.
  • Evidence a standing claim cited has changed, vanished, or come loose from a rewritten commit.
  • A claim has been settled and should be resolved held, failed or superseded.

Do Not Use

  • Independently rechecking a verdict another pass already accepted - that is godmode-second-look.
  • Root-causing a crash or a red suite - that is godmode-investigation.
  • Merging changelog fragments or building a version's release notes - that is godmode-changelog.

Deterministic Execution Flow

  1. Reflect before recording. godmode reflect "<the claim text>" - checks the sentence against what the record already says, and names a contradicting or superseding record before a second one is written.
  2. Repair what already came loose. godmode claim --stale - grounded claims whose cited file evidence changed or vanished since they were recorded; godmode freshness - standing sources that are no longer what was graded; godmode reanchor - citations whose files changed or whose commits are gone.
    • Fallback: before a history rewrite, godmode reanchor --snapshot fingerprints every cited commit; after it, godmode reanchor --remap finds each one's new sha.
  3. Shape the cite to verify first time. godmode claim "<claim>" --grade verified --cite cmd:"<an exit-bearing command>" --verify - every cmd: citation runs through the attested checker first. Exit-bearing forms (git diff --quiet, grep -q, cmp -s, test, git merge-base --is-ancestor) verify directly; a state-reporting command (git status, ls) caps at observed no matter what it prints. Name a file line with file:<path>#L<n> and a record with seq:<n>.
    • Fallback: a claim that still downgrades names why in next_action; fix that citation, never wrap the command in a script to change its grade.
  4. Give a hypothesis its falsifier. godmode claim "<hypothesis>" --grade hypothesis --refuted-by "<the one command or observation that would refute it>".
  5. Run the falsifiers that aged. godmode verify --falsifiers --dry-run lists every hypothesis or incident falsifier aged past two days with no attestation; then godmode verify --falsifiers runs each and attests it on its own.
  6. Resolve once, with fresh evidence. godmode claim --resolve <seq> --outcome <held|failed|superseded> --cite <evidence the claim did not already cite> - a claim resolves at most once.
  7. Clear the Stop-gate backlog. godmode status remaining --digest - its claims.parked count is the claim-shaped sentences the Stop gate parked unrecorded; each gets a claim from step 3 or is softened. For the public surfaces, godmode claim --scan lists claim-shaped sentences whose line names no reproduction and no claim record.

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

Must Not

  • Never write a wrapper script to turn a state-reporting command into an exit-bearing one; use an exit-bearing form directly.
  • Never record a hypothesis without --refuted-by; instead name the observation that would refute it, since a theory nothing could refute is not a finding.
  • Never re-assert a stale claim over its drifted evidence; repair it from step 2 or resolve it superseded.
  • Never resolve a claim with only the evidence it was recorded with, or resolve one sequence twice; instead cite fresh evidence once, and supersede if the outcome later changes.
  • Never run verify --falsifiers before its --dry-run listing; know what will run before it attests.

Acceptance

  • A claim citing an exit-bearing command through --verify records at the grade it asked for.
  • godmode verify --falsifiers --dry-run lists no falsifier aged past due once step 5 has run.

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.