godmode-governance
Preview and govern protected engineering actions without executing them implicitly.
Skill by AIimagined
Chalk Labs
Need help deploying godmode-governance?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/aiimagined-godmode-governance
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: backend
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/godmode-governance
What's inside
- Skill name
godmode-governance- Description
- Preview and govern protected engineering actions without executing them implicitly. Use before destructive, externally visible, history-changing, database-changing, release, credential, branch, or worktree operations. Not for reads, tests or edits inside the working tree.
- License
- Apache-2.0
- Instructions
- 88 lines in SKILL.md · view on GitHub
Files (5)
- PURPOSE.md
- SKILL.md
- openai.yaml · agents/
- godmode-evals.json
- godmode-protection-matrix.md · references/
SKILL.md (first lines)
# Godmode Governance
## Outcome
Turn a risky requested operation into an explicit, reviewable contract. This skill classifies and previews the operation; it does not grant the host permission or claim to intercept every tool.
## Workflow
1. Inspect the current project identity, branch or worktree, dirty state, and the exact target.
2. Separate observation from mutation. Read-only inspection needs no capability.
3. Classify the proposed action with:
```powershell
<plugin-root>/bin/godmode --project <path> guard --operation "<exact operation>"
```
4. For a protected result, run `godmode guard --operation "<exact command>"` and present its `brief` (Context / Options / Resolution / Accepted cost) verbatim.
5. If authorization is required, configure the local authorization secret once and issue a short-lived, one-use capability for the exact action. Never store or pass the secret through Godmode records.
6. Execute only when the user has authorized the mutation and the host provides an appropriate execution boundary. Keep execution separate from classification.
7. Consume the matching capability immediately before the protected operation, then verify the result and record evidence.
Read [godmode-protection-matrix.md](references/godmode-protection-matrix.md) when deciding whether an operation is protected or when writing the preview.
## Fail-closed rules
- Treat an unknown mutation as protected.
- Preserve unrelated user changes and existing worktrees.
- Rewrite history, remove data, publish, install, release, or change a database only from an explicit instruction, never an inferred intention.
- Reject a capability whose action, project, expiry, nonce, or signature does not match.
- Represent a preview as a preview and a guard result as advisory - neither is approval or automatic enforcement.
- Require a rollback or recovery statement for schema, history, release, and destructive filesystem changes.
## Gate routing
- `environment --target X` classifies blast radius before any mutation; unknown fails closed as production and repository text cannot re-label it.
- `egress --staged` scans staged and untracked content for secret shapes before a commit.
- `db --propose` walks the schema ladder: existing column, existing table, and only then a reviewed new table.
- `planmode specify|start|approve|check|arbitrate|bind` gates mutation behind a spec-backed approved plan; `rewind --to SEQ` previews a rollback to a verified checkpoint.
- `ceilings --spent ...` stops a run that exceeded its declared budget; `removal record|why` keeps deletions explicable.
- `godmode ownership --check [--diff-only]` shows which gate rule owns each path or command, from the classifier's own tables, and fails closed when the checked-in decision table no longer matches the classifier it was generated from - run it before a commit that touches gate-owning paths.
## Absorption verdicts
An upstream item (a dependency bump, a vendor release note, a competitor's
fix) gets a `decision` record with subject `absorb:<item>` and BOTH verdicts
in its data: `import_verdict` (adopt | extend | diverge | skip | exists |
unread | n-a) and `behaviour_verdict` (confirmed-have | confirmed-dont |
unverified). `confirmed-*` needs a `file:` citation proving it. "n-a -
different surface" answers whether we can import it, never whether the same
defect lives in our own implementation - that is what the behaviour verdict
is for, and an item with only an import verdict is half-recorded. `unread`
is an honest import verdict but is never a settled one: it names a surface
read that never opened the source, so it still leaves the item
half-recorded until a real verdict replaces it.
`remember --kind decision --subject "absorb:<item>" --value "import_verdict:
<verdict>. behaviour_verdict: <verdict>."` validates both verdicts and
refuses an adopt or extend verdict that cites no `file:<path>` source
(README, docs, and release notes do not count as a source cite); a
surface read may still say unread, skip, or diverge. Once a write passes,
the parsed verdicts are persisted onto the record's data, so
`godmode_runtime.godmode_parity.upstream_verdicts(archive, items)` reads a
CLI-recorded decision the same way it reads one written by direct archive
access (`archive.append("decision", "absorb:<item>", {"import_verdict": ...,
"behaviour_verdict": ...}, evidence=["file:..."])`), and reports
`verdicted` / `half_verdicted` / `unread` per item.
## Completion
Report the classification, whether authorization was required, what actually ran, the recovery boundary, and fresh verification. If execution was outside the available host boundary, stop after the preview and say so plainly.
## Verbs at the moment of demand
- Before declaring the work complete: the Stop gate blocks once when this session's asks, the plan's pending steps, or uncited criteria are still open on the record; close each with the command its line shows, or say which are deferred and why.
- Before a docs or release pass that must carry nothing personal: `godmode privacy --repo` names every email, home path, host-positioned IP address, secret shape and oversized tracked file by path and line, value masked.
- Before anything leaves the machine: `godmode precheck --preflight` (suite, scans, census, reach on a disposable worktree); `godmode egress --staged` says exactly what a push would send; `godmode privacy` and `godmode untrusted` scan for what must not leave; `godmode netgate` audits what dialled out.
- Before any push: `godmode precheck --preflight --suite-shards 4` validates HEAD in a disposable worktree - the designated suite, the workflow's own gate steps read from `.github/workflows/godmode-verify.yml`, the banned-term scan - and records a `preflight` attestation at that commit. `godmode authorize stage` refuses to stage a push or a release without a green one at HEAD; `--without-preflight "<reason>"` is recorded, never silent.
- Cutting a release: `godmode changelog merge --set-version X`, `godmode release-notes build X` derives the note from that section and `godmode release-notes check X` holds it to the shape (present, every entry covered, no empty section, no process narration, a Verifying section), `godmode version --reconcile`, `godmode bindings --write`, `godmode release` compares two releases, `godmode checksums` and `godmode sbom` record what shipped, `godmode license` classifies what was read.
- After a source edit: `godmode retest` lists every test that pins a changed file (names its path, module, or stem) as one command per runner; `godmode retest --run` executes it and attests the exit code, so the done bar can cite `retest` instead of a hand-picked test.
- Before closing a session on a project that declares debt counters (`.godmode-ratchets.json`, name to command): `godmode ratchet run` records each value and names every counter that rose; `integrity` carries the same finding.
- Held-back checks: `godmode oracle hold --command "<check>"` (operator password) keeps a check under the git metadata directory where the agent never chose it; `godmode oracle list` and `godmode oracle run` read and run them, and `claim --verify` runs them after the cited checks - a red one caps the grade at observed.
- A protected operation: `godmode authorize stage --operation "<exact command>"` after the operator's password, `godmode stage` inspects what is staged, `godmode capabilities` and `godmode approvals` list what stands, `godmode ceilings` sets spend bounds, `godmode protect` pins a file, `godmode fence` and `godmode scope` bound what an edit may touch, `godmode boundaries` shows the host's enforcement table.
- A rule to keep or a decision to weigh: `godmode register` pins an evaluator, `godmode precedent` records how a prior case was decided, `godmode recurring` schedules a standing check, `godmode sop` records a procedure, `godmode arbitrate` settles two conflicting records, `godmode governance` reports the posture.
- Undoing or removing: `godmode rollback` and `godmode rewind` restore a recorded state, `godmode removal` records a deliberate deletion, `godmode expunge --sequence N --reason` retracts one record, `godmode branches` lists what diverged.
- Configuration and environment: `godmode config check`, `godmode operator --policy` (which layer decided each gate), `godmode environment`, `godmode locale check`, `godmode extensions`, `godmode paired-artifact` (files that must stay in sync), `godmode parity` and `godmode upstream` against the reference repository (`upstream --path <tree> --skills <keyword>` lists its skill docs that touch a keyword, the parity read before a verdict), `godmode freshness` for research that aged.
- Measuring the project: `godmode metrics`, `godmode metric-contract`, `godmode trends`, `godmode minimality`, `godmode assurance`, `godmode trust`, `godmode selftest`.From AIimagined/Godmode, Apache-2.0.
More from AIimagined/Godmode
- 0.627score
godmode-continuity
AIimagined
Reconstruct and preserve private project continuity from local evidence.
SkillAFreshness: A★ 6Claude Code +9 - 0.627score
godmode-investigation
AIimagined
Diagnose technical failures with reproducible evidence and bounded hypotheses.
SkillAFreshness: A★ 6Claude Code +9 - 0.627score
godmode-skill-forge
AIimagined
Create an original, compact project skill after a repeated capability gap is proven.
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-second-look
AIimagined
Independently recheck a change, session, PR, or diff that already carries an accepted claim or verdict.
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-skill-eval
AIimagined
Score a drafted or hand-edited skill's authored routing rows against the committed baseline, and confirm its bundle validates, before it ships or is enabled…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-changelog
AIimagined
Draft and verify release-notes from local commits and changelog fragments.
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-ci
AIimagined
Run CI's checks locally, in CI's order, before opening a pull request: the task precheck, the fragment gate against the base branch, the preflight of HEAD or…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-codegraph
AIimagined
Query a local, on-demand code dependency graph to answer who calls, who imports, and what blast radius a change carries, before spending an unnecessary full…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-evidence
AIimagined
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…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-host-sync
AIimagined
Detect every installed AI CLI host, reconcile Godmode's interception manifests into each one's native format, and drift-repair the installed configuration…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-spec-lifecycle
AIimagined
Move a specification and its plan through specify, approve, implement, review, and status as one linked, resumable record instead of separate hand-authored…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-impact-gate
AIimagined
Score a staged commit's fan-out and architectural-boundary crossings just before committing: dependents left unfollowed, hook-to-runtime import direction…
SkillAFreshness: A★ 6Claude Code +9
Signals
- Stars
- 6
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 25, 2026 (1 month ago)
- Last release
- Oct 5, 2026 (29 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- Apache-2.0
- Repository
- github.com/AIimagined/Godmode
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from AIimagined
- 0.627score
godmode-skill-forge
AIimagined
Create an original, compact project skill after a repeated capability gap is proven.
SkillAFreshness: A★ 6Claude Code +9 - 0.627score
godmode-investigation
AIimagined
Diagnose technical failures with reproducible evidence and bounded hypotheses.
SkillAFreshness: A★ 6Claude Code +9 - 0.627score
godmode-continuity
AIimagined
Reconstruct and preserve private project continuity from local evidence.
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-research
AIimagined
Study an outside source repository in a fixed order - licence, tree, the two or three implementing source files, line-cited receipts - and record both an…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-skill-eval
AIimagined
Score a drafted or hand-edited skill's authored routing rows against the committed baseline, and confirm its bundle validates, before it ships or is enabled…
SkillAFreshness: A★ 6Claude Code +9 - 0.622score
godmode-second-look
AIimagined
Independently recheck a change, session, PR, or diff that already carries an accepted claim or verdict.
SkillAFreshness: A★ 6Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get godmode-governance running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
godmode-governance skill. Agent Mart. https://agentmart.dev/skills/aiimagined-godmode-governance (last updated 2026-10-24)