What problem does it solve? After building an endpoint or command, you need proof that its contract actually holds against the live target, not just a claim in a plan. This Skill runs a real probe against the API or CLI, checks status codes, exit codes, and payloads, and records the outcome as anchored manual evidence the DoD gate can re-read. ## Core Features & Use Cases - API Probe: Call a real HTTP endpoint and assert its contract via status code and payload fields. - CLI Probe: Run a target command and assert its contract via exit code and stdout/stderr pattern. - Anchored Evidence: Write each probe result as a manual check with evidence_tier: anchored and an anchor_id, reusing the existing manual-test admissibility tier. - Safety Model: Probes are never gate-driven, run only on deliberate invocation, and require human/agent review before any command hits the shell. - Use Case: After shipping a new REST endpoint, invoke the API probe to confirm it returns the expected status and payload fields, then anchor the passing result so the DoD gate accepts it as verified evidence. ## Quick Start Ask the agent to run a contract-test API probe against your new endpoint and anchor the passing result as manual evidence.
