ai-review-evidence-reconciler

Reconcile AI-generated review evidence against expected verification criteria in read-only mode.

Updated Sep 16, 2026
One-click install
npx skills add https://github.com/Military-Veteran-Team-LPT-Realty/mvt-manus-public-skills --skill ai-review-evidence-reconciler-military-veteran-team-lpt-realty
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ai-review-evidence-reconciler
Source: https://github.com/Military-Veteran-Team-LPT-Realty/mvt-manus-public-skills/tree/main/skills/ai-review-evidence-reconciler
Command: npx skills add https://github.com/Military-Veteran-Team-LPT-Realty/mvt-manus-public-skills --skill ai-review-evidence-reconciler-military-veteran-team-lpt-realty

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-generated code reviews and test summaries can contain claims that contradict or lack support from actual execution logs. This Skill provides a systematic, fail-closed workflow to cross-check those claims against raw artifacts before any approval decision is made. ## Core Features & Use Cases - Evidence Reconciliation: Compare AI review claims line-by-line against test logs, build outputs, and baseline criteria, classifying each finding as Match, Discrepancy, or Unverified. - Fail-Closed Verification: Missing logs, ambiguous evidence, or unverified assertions are flagged as verification failures by default. - Standardized Reporting: Generate a reconciliation report from the included template with confidence ratings, risk levels, and an explicit human approval gate. - Use Case: Before merging a release candidate, run this Skill to audit the AI-generated review summary against the actual CI test logs and produce a signed-off reconciliation report for reviewers. ## Quick Start Reconcile the AI review summary for PR #123 against the test execution logs and generate a reconciliation report using the evidence template.

Frequently Asked Questions about ai-review-evidence-reconciler

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify AI-generated code review claims against test logs?

Gather the AI review summary, test execution logs, and build artifacts into a review scope, then compare each claim line-by-line against the raw logs. Classify findings as Match, Discrepancy, or Unverified and aggregate them into a confidence rating.

What is fail-closed verification in evidence reconciliation?

Fail-closed verification means any discrepancy, missing log, unverified assertion, or ambiguous evidence is flagged as a verification failure by default. Claims are only accepted when directly supported by underlying logs and criteria.

Can this Skill modify code or approve deployments automatically?

No. All analysis operates strictly in read-only mode and never modifies source code, test artifacts, or production systems. Reports may recommend actions, but execution requires explicit human authorization and separate sign-off.

What artifacts are needed to run an evidence reconciliation audit?

You need the AI review summaries, test execution logs, build outputs, compliance checklists, and the canonical expected criteria or acceptance thresholds. File integrity and read-only access should be confirmed before comparison begins.

Does the reconciliation report include secrets or credentials?

No. The workflow enforces a zero-secret policy, so API keys, passwords, authentication tokens, and personal identifiers are never included in review evidence summaries or logs.