stretch-testing
Skill4 starsChanged 8 months agoMITView on GitHub
Adversarial QA and BRD compliance testing — write failing tests that prove bugs, edge case probing, and spec-vs-implementation mismatch detection. Use when running stretch tests, QA audits, or BRD compliance checks.
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- cli-pipelineSkill
- intake-engineSkill
- template-engineSkill
- skill-creatorSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Before writing any test, read the relevant BRD section to understand expected behavior:
- Distinctive: line 57
Proof: test file path + test name
- Very specific: line 63
Import helpers from tests/templates/helpers.ts
- Distinctive: line 67
Write tests without BRD reference for expected behavior
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Stretch Testing Skill
When to Use
- Running adversarial QA against the codebase
- Detecting BRD spec vs implementation mismatches
- Writing edge case tests for untested code paths
- Verifying state machine can't be bypassed
- Checking template outputs match BRD §9 schemas exactly
Methodology
1. Read Authority First
Before writing any test, read the relevant BRD section to understand expected behavior:
- State machine: BRD §6
- CLI commands: BRD §7
- Intake modes: BRD §8
- Output schemas: BRD §9
- Pack matrix: BRD §10
- Patch rules: BRD §12
- Promote criteria: BRD §13
2. Adversarial Test Categories
| Category | What to Test | Example |
| BRD Mismatch | Implementation differs from BRD spec | Budget values hardcoded vs BRD §12.4 |
| State Bypass | Command runs in wrong state | run working in MIGRATING |
| Input Injection | Dangerous names, paths, content | Project name CON, ../../../etc/passwd |
| Budget Overflow | Word budget can be exceeded | Patch that inflates beyond 1800 words |
| Schema Violation | Generated output missing required fields | Config without schemaVersion |
| Gate Bypass | Promote/completeness gate insufficient | Promote with unresolved questions |
| Compile Drift | Compile output inconsistent across targets | Cursor missing role that Claude has |
3. Test Output Format
it('B-XXX: <description of what should happen per BRD>', () => {
// Arrange: set up the scenario
// Act: trigger the behavior
// Assert: verify BRD-expected result
});4. Reporting Format
For each bug found:
- Bug ID: B-001, B-002, etc.
- Severity: CRITICAL / HIGH / MEDIUM / LOW
- Component: which
src/module - Actual: what happens
- Expected: what BRD says should happen (cite section)
- Proof: test file path + test name
Test Files
- Adversarial tests go in
tests/adversarial/ - Name pattern:
stretch-<category>.test.ts - Import helpers from
tests/templates/helpers.ts
Do NOT
- Fix bugs — only find and document them
- Write tests without BRD reference for expected behavior
- Skip cleanup in afterEach (temp dirs, external files)
- Assume anything works until proven with a test
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

