skills.rest
AMC-JTC
AMC-JTCCOMMUNITY
Explore all skills by this author →
GitHub

testing

Validates skill conformance and runs tiered test suites with regression classification.

Updated Aug 17, 2026
Search Tags:#testing#validation#health check#regression#test suite#ci#conformance
One-click install
npx skills add https://github.com/AMC-JTC/gbrain-1 --skill testing-amc-jtc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: testing
Source: https://github.com/AMC-JTC/gbrain-1/tree/main/skills/testing
Command: npx skills add https://github.com/AMC-JTC/gbrain-1 --skill testing-amc-jtc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Keeping a growing skill library and codebase healthy requires two kinds of checks that are easy to skip: validating that every skill meets the conformance bar (frontmatter, manifest, resolver coverage) and running the project test suite with enough context to tell real regressions from stale tests and flakes. This Skill automates both. ## Core Features & Use Cases - Skill conformance validation: Verifies every skill directory has a SKILL.md with valid YAML frontmatter, required sections, manifest.json coverage, RESOLVER.md round-trips, and no duplicate triggers across skills. - Tiered test execution: Runs unit tests, evals, integration tests, and system health checks (disk, memory, CPU, service liveness) in structured phases. - Regression intelligence: Uses git history to classify each failure as REGRESSION, STALE, FLAKE, NEW, or INFRA, with a safe auto-fix protocol for stale tests and strict escalation for security failures. - Use Case: A daily cron fires, runs bun test and system health checks, diffs the last 24 hours of commits, and reports that one failing test is a regression caused by a specific commit while two others are known flakes. ## Quick Start Ask the agent to run the tests and tell you what's broken, and it will execute the tiered suite and return a regression-classified report.

Frequently Asked Questions about testing

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

FAQPage Schema
How do I validate that all my skills conform to the SKILL.md standard?▼

Run the conformance mode, which walks every skill directory, parses YAML frontmatter for name and description, checks required sections, and verifies manifest.json and RESOLVER.md coverage round-trip. The automated check is bun test against the skills-conformance, resolver, and plugin manifest test files.

How do I tell if a failing test is a real regression or a stale test?▼

Check git history for each failure: if the tested code changed recently, it is a likely regression; if only the test's expectations are outdated, it is stale. The skill classifies failures as REGRESSION, STALE, FLAKE, NEW, or INFRA and reports the commit that broke each regression.

Can this run as a daily automated test health check?▼

Yes, it is designed for daily cron execution. It runs unit tests, evals, and system health checks, then stores results in ~/.gbrain/test-state.json for trend tracking, flake detection, and regression velocity measurement.

What test failures should never be auto-fixed?▼

Security test failures must always be escalated, never auto-fixed. Behavior mismatches where the test expects A but code does B require asking first, since the code may contain the bug. Skipped tests with TODOs should not be un-skipped without understanding why.

What are the limitations of the regression classification?▼

Classification depends on git history being available and commits being granular enough to attribute failures. Flakes caused by external API timeouts or LLM variance can only be flagged and retried, not eliminated, and ambiguous cases still require user judgment.

Related Skills

Explore More Software Engineering Skills→
garrytangarrytan
29.4k

testing

Validates skill conformance and runs tiered test suites with regression classification.

Community
Advanced
darkmatterdarkmatter
1

caveman-commit

Generate terse Conventional Commit messages from staged diffs.

Official
Basic
huuhiep0303huuhiep0303

doubt-driven-development

Review code decisions adversarially to identify potential issues before deployment.

Community
Advanced
AraszArasz
2

test-economy

Counts full-suite test runs per session and nudges when they exceed the budget.

Community
Advanced