What problem does it solve? When a named human must sign off a readiness stage gate in the Seshat BI pipeline, the evidence they need is scattered across readiness docs, readiness-status.yaml files, metric contracts, and parked-on dependency maps. This Skill composes those committed artifacts into one ordered, reviewable pack without inventing content or granting the approval itself. ## Core Features & Use Cases - Evidence Composition: Collects gate requirements, four-status readiness state, open blockers, unresolved AL1 assumptions, blocking parked-on edges, and pending metric contracts into a single pack at mappings/<table>/approval-evidence-pack-<stage>.md. - Approval Boundary Enforcement: Ends with an empty approval slot (Form A), a read-only recorded approval (Form B), or a not-applicable statement for mechanical gates (Form C); it never writes approvals[] or moves a stage to pass. - Honest-State Handling: Missing or unreadable sources become named blockers rather than fabricated content, and no numeric score or completeness count is ever emitted. - Use Case: Before a metric owner signs the Semantic Model Ready gate for a retail table, run this Skill to produce the full pre-approval pack they review, then hand the decision to the Approval Console. ## Quick Start Ask the agent to assemble the approval evidence pack for table retail_store_sales at the semantic_model_ready stage.
