ship
The evidence and audits support release.
The differentiator
Generative AI made the first draft cheap. It did not make client-ready work automatic. revüe separates model judgment from executable rules: the model can critique hierarchy and propose fixes, while local validators check colors, fonts, claims, hard noes, required structure, audit consistency, and the evidence floor.
ship
The evidence and audits support release.
ship with changes
Ready after named, bounded corrections.
caution
Useful work, but proof or approvals remain thin.
block
A safety, brand, fabrication, or evidence gate failed.
The repository ships with its tests, red-team fixtures, validators, and examples. These are reproducible artifacts—not adoption claims.
Claude Code
/plugin marketplace add gcorrist66/revue-proof-workflow-skill
/plugin install revue-proof-workflow@revue-proof-workflowThe same portable skill layout also runs in Codex and Cowork. The validators use Python's standard library and operate on local files without network or credential access.
No. It prevents unapproved claims from receiving a ship verdict when the claim lock and audit are in use, and it keeps missing proof visible instead of polishing uncertainty into fact.
It is a portable Agent Skill with a workflow, references, schemas, local validators, worked examples, and an automated evaluation suite.
The bundled deterministic validators use Python's standard library, require no credentials or network access, and operate on local files.
Consultants, agencies, internal creative teams, and developers producing stakeholder-facing work with AI who need a repeatable readiness decision before release.
Corriston Consulting builds the evidence floors, brand locks, approval gates, and handoffs that let teams move faster without lowering the bar.