skills.rest
33may
33mayCOMMUNITY
Explore all skills by this author →
GitHub

Verification & Quality Assurance

Validates code quality with truth scoring, CI regression guards, and automatic rollback.

1|1|Updated Nov 28, 2025
Search Tags:#code quality#ci/cd#rollback#verification#github actions#truth scoring#regression testing
One-click install
npx skills add https://github.com/33may/robotics --skill verification-quality-assurance-33may
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Verification & Quality Assurance
Source: https://github.com/33may/robotics/tree/main/humanoid/.claude/skills/verification-quality
Command: npx skills add https://github.com/33may/robotics --skill verification-quality-assurance-33may

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @noble/ed25519.

What problem does it solve? Agent-generated code and rapid changes can silently introduce regressions, security issues, and quality degradation. This Skill provides a verification layer that scores output reliability, runs CI regression guards, and rolls back changes that fail a configurable quality threshold. ## Core Features & Use Cases - Truth Scoring: Computes 0.0-1.0 reliability metrics for code, agents, and tasks with trends, statistics, and export to JSON, CSV, or HTML. - CI Regression Guards: Ships behavioral smoke tests, a tool-description discoverability audit with monotone-decreasing baselines, and Ed25519-signed witness manifests that verify documented fixes remain present in builds. - Automatic Rollback: Reverts changes that fail verification against a configurable threshold (default 0.95), with selective, dry-run, and backup-first modes. - Use Case: A team running multi-agent code generation adds the verification check to their GitHub Actions pipeline so any commit scoring below 0.95 fails the build and is rolled back before merge. ## Quick Start Ask the agent to run a verification check on the current directory with the default 0.95 threshold and show the truth score report.

Frequently Asked Questions about Verification & Quality Assurance

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

FAQPage Schema
How do I run a code verification check with a quality threshold?▼

Run npx ruflo@alpha verify check to verify the current directory, or pass --file or --directory to target specific paths. Use --threshold to set the minimum score, for example --threshold 0.98 for stricter gating.

How do I integrate verification scoring into GitHub Actions CI?▼

Run npx ruflo@alpha verify check --json in a workflow step, parse the overallScore field with jq, and exit with a non-zero code when the score falls below your threshold. The command returns exit code 1 automatically when verification fails.

What is a truth score and how is it interpreted?▼

A truth score is a 0.0-1.0 reliability metric for code, agents, or tasks. Scores of 0.95 and above are production-ready, 0.85-0.94 are acceptable, 0.75-0.84 need attention, and below 0.75 require immediate action.

Does automatic rollback require git?▼

Yes, rollback features require a git repository since reversion is git-based. Git-based rollback completes in under one second, and selective file rollback with automatic backup is also supported.

Why does the witness verification fail in CI?▼

Witness verification fails when a load-bearing marker for a documented fix is missing from the built dist output. Regenerate the manifest with node scripts/regen-witness.mjs after confirming the fix is still present, and ensure @noble/ed25519 is installed.

What are the limitations of the truth scoring dashboard?▼

The documentation notes the truth-scoring, auto-rollback, and WebSocket dashboard surface is only partly shipped, while the CI guards (smoke tests, discoverability audit, witness manifests) are fully operational. Treat the CI Guards section as the authoritative current state.

Related Skills

Explore More Software Engineering Skills→
SPeeDoA1SPeeDoA1
6

code-review

Analyze code changes and generate structured review feedback reports.

Community
Basic
MeoBakaMeoBaka

datadog-automation

Automate Datadog monitoring operations via Rube MCP toolkit.

Community
Advanced
codeApe-7codeApe-7
17

security-reviewer

Analyzes codebases for vulnerabilities, misconfigurations, and secret leaks.

Community
Advanced
BURHANDEV-ENTERPRISEBURHANDEV-ENTERPRISE

Verification & Quality Assurance

Run verification checks with truth scoring and automated rollbacks.

Official
Advanced