Audit AI agent task outputs against project contracts and checklists.
npx skills add https://github.com/XinAloha/skills --skill verify-loop-delivery
Please help me install this Agent Skill. Skill: verify-loop-delivery Source: https://github.com/XinAloha/skills/tree/main/loop-engineering/verify-loop-delivery Command: npx skills add https://github.com/XinAloha/skills --skill verify-loop-delivery
This Skill solves the "self-approval" trap in AI agent loops, where agents tend to overestimate their own success, leading to hidden technical debt and failed deployments. It enforces a strict separation of duties between the implementer and the verifier.
Use the verify-loop-delivery skill to independently audit the current implementation against the project contract and checklist.
High-intent search queries and answers about installing and using this skill.
To prevent AI agent self-approval, you need independent verification that enforces a strict separation of duties between the implementer and the verifier, requiring evidence-based auditing of task outputs.
Verify AI agent task outputs by independently auditing implementation evidence, such as test results and logs, against predefined project contracts and checklists to ensure objective pass, reject, or escalate decisions.
Evidence-based auditing for automated software engineering workflows is the process of requiring verifiable proof like static analysis and regression test results rather than subjective claims to validate code quality and governance.
Yes, applying independent verification during complex refactoring loops ensures that agent-generated code meets all project standards and passes regression tests before it is merged into the main branch.
AI agents overestimate task success and cause hidden technical debt due to the self-approval trap, which is solved by enforcing a strict separation of duties and objective evidence-based checks in the loop.
You need independent verification in AI agent loops when your automated software engineering workflows require strict quality control, non-repudiation, verifiable testing, and objective pass or reject decision-making.
samjhechtGenerate Constitution, Roadmap, and process documentation templates for project governance.
harkersReview bounded work for correctness and output evidence-linked findings with a disposition.
cxnaiveAutomate pre-commit code verification with static security analysis and regression checks.
jayhjenkinsAutomate three-tier QA testing with TEST_INDEX.md and RUNS/ reports.
