What problem does it solve? It prevents uncontrolled implementation work where AI-assisted changes silently expand scope, overwrite working baselines, skip verification, or claim success without evidence, leaving no way to review or roll back what was done. ## Core Features & Use Cases - Bounded Execution Contract: Confirms objective, protected state, authorized scope, acceptance criteria, and recovery path before any consequential change is made. - Checkpoint Verification & Partial-Actuation Containment: Verifies each meaningful change boundary and cleans up or preserves ownership of partially created resources when later gates fail. - Evidence-Backed Closure: Reports exact status (CANDIDATE PASS, ACCEPTED, PARTIAL, BLOCKED, REVERTED, STOP) with diffs, test results, and unverified gates stated honestly. - Use Case: You have an approved plan to modify a parser in a working codebase. This skill implements only the authorized change on an isolated branch, runs fixture tests at each checkpoint, preserves the known-good baseline, and closes with a full evidence report. ## Quick Start Ask the AI to implement the approved plan for your task using signalproof-build, preserving the current baseline and reporting verification evidence for each change.
