Ratify
Memory-safe engineering-peer plugin for Claude Code + Codex: 11 skills, 6 bounded agents, 4 hooks. Claims only what it can prove; closes every turn with one next action.
npx -y skills add menot-you/n-soul --skill ratifyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 15 days oldThe repository was created 15 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Turn an accepted direction into a locked decision and result contract with authority, proof, rollback, and stop conditions.
SKILL.md
2.4 KB, 452 tokens by cl100k_base, as published. Nobody here has run it
Ratify
Use after direction is chosen but before planning or autonomous actions. Ratification removes ambiguity; it does not add implementation detail. Collapse the contract for the engineer into result, proof, authority, rollback, and the first blocked decision.
Inputs to lock
- Accepted direction and the reason it won.
- Observable result.
- Scope in and explicit scope out.
- Non-negotiable constraints.
- Reversible authority already granted.
- Actions that require the engineer's approval.
- Independent verification required for acceptance.
- Rollback path and stop conditions.
Method
- Read the decision source and current repository facts.
- Rewrite the result as an observable state, not an activity.
- Enumerate all capabilities named by the engineer. Put each one in scope or out with a reason, without exposing skills or agents as choices.
- Separate preparation from the action. Mark destructive, public, production, financial, credential, data-loss, and merge-to-main actions as gated unless authority is explicit.
- Apply the closed
witnessandevidencemeanings indecisions.md: name the independent observer and required observation separately from the evidence and independent verification required for acceptance. Apply the no-kernel boundary inSPEC.md§3; do not trust the implementation's own claim. - Define rollback in operational terms and state any irreversible portion.
- Define stop conditions: missing authority, unexpected reach, failed required observation or independent verification, unsafe rollback, or conflict with current state.
- Check correctness, reliability, maintainability, reversibility, security, and reach.
- Return one ratified contract or one blocking question. Never return a menu, multi-question survey, or process ritual as the product surface.
Output
Produce a compact record with:
- Decision and rationale.
- Result.
- Scope in / scope out.
- Constraints.
- Authority and gated actions.
- Witness and required observation, or their exact gap.
- Evidence and independent verification.
- Rollback.
- Stop conditions.
- Known trade-off.
The record is ready when a cold planner can build work units without inventing a product or safety decision.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.