What problem does it solve? After validation and tests pass, teams still need evidence-based judgment on whether a feature and its end-to-end slice actually meet acceptance criteria, plus deterministic classification of every failed criterion. ## Core Features & Use Cases - Evidence-Based Evaluation: Grades each acceptance criterion as pass, fail, or inconclusive using GUI, TUI, API, service, observability, or manual review routes, with evidence recorded in a structured template. - Bounded Observability: Uses Grafana-backed telemetry only for predefined operational criteria with a complete query-admission record, signal provenance, and strict query budgets. - Normalized Failure Events: Emits an immutable failure-event-v1 per failed criterion via scripts/failure_contract.py, binding expected/observed behavior, evidence digests, commit, and spec fingerprints, then routes to proven-owner or diagnosis-required dispositions. - Use Case: Within an Agentic SDLC run, after tests pass for FEAT-001, evaluate its acceptance criteria against the locked plan, observe the end-to-end slice, and record a classified failure event for any criterion that fails. ## Quick Start Ask the Agentic SDLC coordinator to run sdlc-evaluate for the active feature after validation and tests pass so each acceptance criterion is graded with recorded evidence.
