qa-investigation-harness
AGENTS.md4 starsChanged 3 months agoMITView on GitHub
This repo has a risk-driven QA investigation harness. Full rules live in qa-harness/RULES.md — read it in full before running any stage below, and follow it exactly. Read the relevant stage file under qa-harness/stages/ before acting on that stage; do not rely on summaries or memory of it. Never write Playwright test code yourself. Test generation belongs to this project's test-generation agent — see qa-harness/RULES.md for how to get one if it doesn't exist yet (npx playwright init-agents --loop=codex sets…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository6 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- qa-challengeSkill
- qa-exploreSkill
- qa-modelSkill
- qa-validateSkill
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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
- Very specific: line 8
Never write Playwright test code yourself.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
QA Investigation Harness
This repo has a risk-driven QA investigation harness. Full rules live in
qa-harness/RULES.md — read it in full before running any stage below, and
follow it exactly. Read the relevant stage file under qa-harness/stages/
before acting on that stage; do not rely on summaries or memory of it.
Never write Playwright test code yourself. Test generation belongs to this
project's test-generation agent — see qa-harness/RULES.md for how to get
one if it doesn't exist yet (npx playwright init-agents --loop=codex sets
one up for this tool specifically). This harness only investigates, produces
a human-reviewed test plan, and audits the result.
Running a stage
There is no slash command here — when the user asks you to run a stage, read
qa-harness/RULES.md plus the matching file below, then follow it exactly
using whatever ticket ID, test case, or slug the user gave you in their
message. Ask for anything the stage file requires that wasn't given.
| User asks for | Read |
| "explore" / Stage 1 | qa-harness/stages/01-explore.md |
| "model" / Stage 2 | qa-harness/stages/02-model.md |
| "challenge" / Stage 3 | qa-harness/stages/03-challenge.md |
| "validate" / Stage 5 | qa-harness/stages/05-validate.md |
Stage 4 — actually generating tests — is not part of this harness. Point this
project's test-generation agent at the specs/<ticket>.plan.md file Stage 3
produces.
Playwright MCP setup
This harness's explore and validate stages need browser access via Playwright MCP. If it isn't already registered, add it once:
codex mcp add playwright -- npx @playwright/mcp@latest --isolated --caps=core --snapshot-mode=full --timeout-action=30000 --timeout-navigation=30000 --timeout-settle=5000 --test-id-attribute=data-testidThis writes to your global ~/.codex/config.toml, not a file in this repo —
there's nothing to commit for it.
See qa-harness/README.md for the full walkthrough.
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.