CODING AGENT HARNESSES

ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS Prompt 1: Reliability review of a coding agent workflow. Use this to check whether a coding agent followed good practices before you merge its work. [START OF PROMPT] Review this coding agent's workflow log and evaluate whether it followed good reliability practices (tests run before claiming success, narrowly scoped changes, checked for side effects), flagging anything risky: [PASTE LAB OR PRODUCTION CODING-AGENT WORKFLOW LOG/DIFF HERE]. Output format: a practice checklist (Practice, Followed? Y/N, Risk if No) plus an overall merge-safety recommendation. [END OF PROMPT] Prompt 2: Verification gate pipeline. Use this to design the checks that every agent-written change must pass. [START OF PROMPT] Design a verification-gate pipeline (unit tests, linting, type checking, security scanning) that must pass before any agent-authored change merges, tailored to this project's existing tooling: [PASTE LAB OR PRODUCTION PROJECT TOOLING/CI CONFIG HERE]. Output format: a pipeline spec (Check, Tool, Pass Criteria, Failure Behavior). [END OF PROMPT]