harness-runtime-feedback

What problem does it solve? Static checks and tests alone cannot prove that a system behaves correctly at runtime. This Skill audits a project for observable runtime surfaces and produces a concrete evidence plan so validation is grounded in real logs, health checks, and error signals rather than assumptions. ## Core Features & Use Cases - Runtime Surface Audit: Inspects the project for logging, health/readiness endpoints, error handlers, retry and timeout policies, and browser console or network evidence. - Structured Evidence Plan: Outputs a Markdown plan listing exact files, commands, endpoints, and expected failure signals, marking anything missing as unknown instead of inventing observability. - Use Case: Before validating a new API feature, run this Skill to identify which log files to tail, which health endpoint to poll, and which error message confirms a failure, so reviewers can verify behavior with concrete evidence. ## Quick Start Ask the agent to build a runtime evidence plan for the current task, covering logs, health checks, and expected error signals.