confidence-scoring

What problem does it solve?

It prevents AI coding workflows from progressing on artifacts that lack verifiable evidence by scoring outputs against a rubric and enforcing confidence gates.

Core Features & Use Cases

• Rubric-based scoring: Evaluates an artifact across rubric dimensions using evidence-required criteria and produces a numeric score.
• Decision gating: Converts the score into PASS/WARN/BLOCK using configured thresholds to control progression safely.
• Evidence auditing: Returns per-dimension evidence-found plus explicit gaps and remediation steps for missing proof.
• Workflow containment: Ensures durable progress claims are only allowed when required receipts/commands/proofs are present.

Quick Start

Have your AI call the scoring command in chat to score a specific artifact type and path, then use the returned status to decide whether to proceed.