agentleFS
aaronreneaaronrene/knowtation/.cursor/skills/freeze-review/SKILL.md

freeze-review

Skill1 starsChanged yesterdayMITView on GitHub

Use when a phase output is declared frozen: true and downstream phases consume it as ground truth without re-deriving (SD-3 Thinking → Auto boundary; OVERSEER-KIT-SPEC §6). Review frozen specs and contracts before Auto build phases depend on them. Default path is automated review with file + line citations for every finding; human escalation only for security, irreversibility, real money, or Tier-3 gates. For bounded review → fix → re-review until pass, use cursor/skills/freeze-review-loop/SKILL.md (/freeze-review-loop). Not automatic — invoke when a…

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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 when a phase output is declared frozen: true and downstream phases consume it as ground truth without re-deriving (SD-3 Thinking → Auto …

    line 3
  2. Distinctive:

    Before Tier 3 action consumes a frozen artifact (merge, staging push, live flip)

    line 22
  3. Very specific:

    Every finding must include path:line so the operator can verify — never trust uncited review output.

    line 70

≈ “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

Freeze-step review skill

Use when a phase output is declared frozen: true and downstream phases consume it as ground truth without re-deriving (SD-3 Thinking → Auto boundary; OVERSEER-KIT-SPEC §6).

Purpose

Review frozen specs and contracts before Auto build phases depend on them. Default path is automated review with file + line citations for every finding; human escalation only for security, irreversibility, real money, or Tier-3 gates.

Read first

  • docs/OVERSEER-KIT-SPEC.md §6 — Freeze-Contract review policy
  • policy/freeze-contract.schema.yaml (K5) — phase output declarations
  • .overseer/config.yaml → freeze_contract.human_escalation

When to run

  • Roadmap step marked Thinking → Auto and {step}a contract artifacts are Muse/Git committed
  • Any artifact listed as frozen: true in a phase freeze block
  • Before Tier 3 action consumes a frozen artifact (merge, staging push, live flip)
  • Ad-hoc / side research via /check-ok (scaffolds docs/reviews/ then this gate)

Review checklist

  1. Ground-truth edge: does a later phase treat this output as truth without re-deriving?
  2. Completeness: interfaces, fail-closed rules, test matrix present?
  3. Security: injection surfaces, secrets, scope leaks?
  4. Irreversibility / real money: migrations, deletes, billing?
  5. Tier 3 linkage: does this artifact gate merge, staging, or live gates?

Verdicts

VerdictMeaning
passNo blocking findings; record review stamp
findingsCited list; resolve or escalate per config
blockedHuman required before downstream Auto build

CLI (when K5 lands)

/path/to/overseer-kit/cli/ok review --freeze <path> --dry-run
/path/to/overseer-kit/cli/ok review --freeze <path>

Multi-round loop (optional)

For bounded review → fix → re-review until pass, use cursor/skills/freeze-review-loop/SKILL.md (/freeze-review-loop). Not automatic — invoke when a frozen spec must clear before {step}b Auto build.

After Auto build (honesty gate)

Freeze review guards the spec. cursor/skills/build-verification-review/SKILL.md (/build-verification-review) guards the implementation — run before marking the phase DONE. Manual/opt-in today; required by SD-3 discipline even when not enforced by CLI.

Escalate to human when

  • security — auth, scope, secrets, injection
  • irreversible — data deletion, non-revertible migrations
  • real_money — billing, live model spend
  • gates_tier3 — artifact gates merge, staging push, or live flip

Everything else: agent resolves or reports without a human stop.

Must cite

Every finding must include path:line so the operator can verify — never trust uncited review output.

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.