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

godmode

Skill6 starsChanged 3 months agoApache-2.0View on GitHub

Coordinate local context continuity, evidence, and guarded coding workflows. Use when starting or resuming substantive repository work, when current reality is uncertain, or when a request spans multiple Godmode capabilities. Not for a single-file edit with no continuity, evidence or gating concern.

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

29 rules. 0 are common, 4 are distinctive, and 25 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 25

    Very specific

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

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

    If the session hooks already delivered a brief (identity, last checkpoint, open obligations, standing laws, the host's enforcement table), …

    line 15
  2. Very specific:

    Run a verb only for what is missing or task-specific: resume --refresh after a compaction, a branch switch or an identity-drift finding.

    line 18
  3. Very specific:

    Use godmode-continuity for session recovery, inventory drift, checkpoints, handoffs, decisions, fixed invariants, or missing context.

    line 36
  4. Very specific:

    Use godmode-investigation for bugs, failures, regressions, evidence gathering, root-cause work, or repeated unsuccessful attempts.

    line 38
  5. Very specific:

    Use godmode-governance before Git history changes, branch/worktree mutation, database changes, releases, deployments, destructive …

    line 40
  6. Very specific:

    Use godmode-repair when the operator says an answer did not land, asks for clarity, repeats a question, or asks what is being waited on.

    line 42
  7. Very specific:

    Use godmode-skill-forge only after a repeated reusable capability gap is proven.

    line 44
  8. Very specific:

    Use godmode-second-look to independently recheck a change, session, PR, or diff that already carries an accepted claim or verdict, before …

    line 45
  9. Very specific:

    Use godmode-changelog to check the changelog-fragment gate, merge fragments for a version, and build and check that version's release notes.

    line 47
  10. Very specific:

    Use godmode-code-of-law at the start of every session and before every new task, when GODMODE-CODE-OF-LAW.md exists at the project root.

    line 49
  11. Very specific:

    Use godmode-host-sync to check whether every installed host still has the hooks wired and to reconcile a drifted or newly detected host's …

    line 51
  12. Very specific:

    Use godmode-codegraph to rebuild or read the local code dependency graph: blast radius of a change, who calls a symbol, which tests a …

    line 53
  13. Very specific:

    Use godmode-skill-eval to score a skill's routing and behaviour against its baseline and to ratchet a regression rather than write it into …

    line 55
  14. Very specific:

    Use godmode-spec-lifecycle to take a spec from draft through review, reconciliation and approval without approving its own plan or …

    line 57
  15. Very specific:

    Use godmode-memory-gardener at session close to list duplicate or contradicting lessons, preview expiry before it runs, and promote a ripe …

    line 59
  16. Very specific:

    Use godmode-impact-gate before committing a multi-module edit: fan-out, import direction, fence and paired artifacts, with a recorded …

    line 61
  17. Very specific:

    Use godmode-research to survey an outside codebase: licence first, receipted source reads, and both verdicts, never adopt or extend from a …

    line 63
  18. Very specific:

    Use godmode-replicate to rebuild a studied mechanism under Godmode names, pinned by a test proven able to fail and cited on its host row.

    line 65
  19. Very specific:

    Use godmode-evidence when a claim keeps being downgraded or cited evidence drifted: exit-bearing cites, falsifiers, stale-citation repair, …

    line 67
  20. Very specific:

    Use godmode-triage to work open asks down to closed, promoted, mapped or parked.

    line 69
  21. Very specific:

    Use godmode-loop-warden to cap a fix-and-retry cycle and cut it off when it repeats.

    line 70
  22. Very specific:

    Use godmode-budget to check a session's or a release's spend against its ceilings.

    line 71
  23. Very specific:

    Use godmode-ci to run CI's checks locally in CI's order before a PR, by shard.

    line 72
  24. Distinctive:

    Do not invoke every specialist.

    line 74
  25. Distinctive:

    Before editing, state the active objective, relevant repository identity, current evidence, unresolved conflict, material assumption, and …

    line 79
  26. Distinctive:

    Before reporting completion, run fresh verification that directly proves the claim.

    line 84
  27. Distinctive:

    Never submit prompts, transcripts, source bodies, credentials, environment dumps, or private instructions to the archive.

    line 90
  28. Very specific:

    Before saying that "the gate blocked" a tool, read hooks status.

    line 97
  29. Very specific:

    Read godmode-command-surface.md only when the requested operation needs exact CLI syntax.

    line 112

“Common” is measured: kept by at least 6 in 10 of the 7 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

Godmode

Outcome

Establish what is true now, choose one bounded workflow, and leave locally verifiable continuity without placing operational memory in tracked project files.

Entry: one path

  1. If the session hooks already delivered a brief (identity, last checkpoint, open obligations, standing laws, the host's enforcement table), that IS the entry: consume it. Do not run context status, resume or session open again for what it gave.
  2. Run a verb only for what is missing or task-specific: resume --refresh after a compaction, a branch switch or an identity-drift finding; session open when no brief arrived (a host whose reach is not HARD, read from hooks status); context status --scan when filesystem drift matters; charter when the project's rules changed.
  3. The CLI is the bundled scripts/godmode.py under the plugin root ($GROK_PLUGIN_ROOT, else $CLAUDE_PLUGIN_ROOT, else two levels above this file), called through <plugin-root>/bin/godmode (bin\godmode.cmd on Windows); it probes python3, python, py and honours GODMODE_PYTHON. Never write a bare python into a command.
  4. Separate observed facts, declared intent, assumptions, conflicts and open obligations, then route to the narrowest specialist below.

On a host whose reach is PARTIAL or SOFT, the path is six verbs: init once, session open, resume, remember, claim, checkpoint; stage a protected operation with authorize stage before it. On Grok a brief rides the next allowed call and a refusal reaches the model after the call; both are advisories.

Routing

  • Use godmode-continuity for session recovery, inventory drift, checkpoints, handoffs, decisions, fixed invariants, or missing context.
  • Use godmode-investigation for bugs, failures, regressions, evidence gathering, root-cause work, or repeated unsuccessful attempts.
  • Use godmode-governance before Git history changes, branch/worktree mutation, database changes, releases, deployments, destructive filesystem work, or external writes.
  • Use godmode-repair when the operator says an answer did not land, asks for clarity, repeats a question, or asks what is being waited on.
  • Use godmode-skill-forge only after a repeated reusable capability gap is proven.
  • Use godmode-second-look to independently recheck a change, session, PR, or diff that already carries an accepted claim or verdict, before it is relied on.
  • Use godmode-changelog to check the changelog-fragment gate, merge fragments for a version, and build and check that version's release notes.
  • Use godmode-code-of-law at the start of every session and before every new task, when GODMODE-CODE-OF-LAW.md exists at the project root.
  • Use godmode-host-sync to check whether every installed host still has the hooks wired and to reconcile a drifted or newly detected host's manifest.
  • Use godmode-codegraph to rebuild or read the local code dependency graph: blast radius of a change, who calls a symbol, which tests a change must retest.
  • Use godmode-skill-eval to score a skill's routing and behaviour against its baseline and to ratchet a regression rather than write it into the snapshot.
  • Use godmode-spec-lifecycle to take a spec from draft through review, reconciliation and approval without approving its own plan or hand-editing the compiled law.
  • Use godmode-memory-gardener at session close to list duplicate or contradicting lessons, preview expiry before it runs, and promote a ripe lesson with a cited rationale.
  • Use godmode-impact-gate before committing a multi-module edit: fan-out, import direction, fence and paired artifacts, with a recorded decision as the escape hatch.
  • Use godmode-research to survey an outside codebase: licence first, receipted source reads, and both verdicts, never adopt or extend from a surface read.
  • Use godmode-replicate to rebuild a studied mechanism under Godmode names, pinned by a test proven able to fail and cited on its host row.
  • Use godmode-evidence when a claim keeps being downgraded or cited evidence drifted: exit-bearing cites, falsifiers, stale-citation repair, one resolution.
  • Use godmode-triage to work open asks down to closed, promoted, mapped or parked.
  • Use godmode-loop-warden to cap a fix-and-retry cycle and cut it off when it repeats.
  • Use godmode-budget to check a session's or a release's spend against its ceilings.
  • Use godmode-ci to run CI's checks locally in CI's order before a PR, by shard.

Do not invoke every specialist. One specialist owns each overlapping capability; use a second only when the request genuinely crosses its boundary.

Working contract

Before editing, state the active objective, relevant repository identity, current evidence, unresolved conflict, material assumption, and next safe action. Inspect before mutating. Preserve unrelated user changes. Keep the implementation surface as small as the complete solution allows.

Before reporting completion, run fresh verification that directly proves the claim. Then record a checkpoint with evidence and remaining obligations. Never convert a passing test, agent report, or plausible inference into a broader claim than it proves.

Privacy boundary

Never submit prompts, transcripts, source bodies, credentials, environment dumps, or private instructions to the archive. Store structured facts, hashes, relative paths, statuses, and evidence references only. Godmode is on demand: do not start a watcher, listener, proxy, daemon, update ping, or background process.

Enforcement honesty

Before saying that "the gate blocked" a tool, read hooks status. HARD means the host called the hook and a live deny was chronicled. PARTIAL, DEGRADED, SOFT or UNAVAILABLE means the host is not proven to call the hook: say that the CLI refused, or that the preview would refuse, and never that the host stopped the tool.

Closing

Close substantive work through session close once. An unattested HARD rule, an uncited claim, a half-done pair, a regression (a step green earlier this session and red since), or a declared perimeter check that never ran (perimeter add|run) blocks it. A check already run on this HEAD this hour is evidence: claim --verify reuses its attestation instead of running it again. config check, roles and operator validate declared configuration; locale check validates translated guidance.

Read godmode-command-surface.md only when the requested operation needs exact CLI syntax. On a host with no godmode hook support, read godmode-generic-adapter.md for the manual wiring path.

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.