proqi / tests
AGENTS.md45 starsChanged 51 days agoMITView on GitHub
These rules apply to the complete first-party integration test tree. - Name each test file for one behavior, contract, or system boundary. The file owns only scenarios and helpers for that boundary. - An integration-test crate root must declare one of two roles in its module-level contract. It is either a registry only, or it owns one cohesive core contract while registering narrower satellite modules. Top-level tests are allowed only in the second form and must belong to that declared…
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 31,439 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 repository11 other files it gives its agents.
- AGENTS.mdAGENTS.md
- xtask/AGENTS.mdAGENTS.md
- architecture-reviewSkill
- claude-code-reviewSkill
- codex-reviewSkill
- implement-in-worktreeSkill
- proqi-screenshotSkill
- releaseSkill
- walkthroughSkill
- proqi-debugSkill
- proqiSkill
The file
We split this AGENTS.md 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 27
Keep setup and assertions beside the behavior they prove when they are used by one module.
- Very specific: line 31
Do not create catch-all modules such as misc, common_tests, or helpers.
- Distinctive: line 37
Test one semantic owner at the narrowest layer that can prove it.
- Distinctive: line 42
Keep failure paths in the same behavior module as the successful contract so their ownership cannot drift.
≈ “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
Integration test ownership
These rules apply to the complete first-party integration test tree.
Suite boundaries
- Name each test file for one behavior, contract, or system boundary. The file owns only scenarios and helpers for that boundary.
- An integration-test crate root must declare one of two roles in its module-level contract. It is either a registry only, or it owns one cohesive core contract while registering narrower satellite modules. Top-level tests are allowed only in the second form and must belong to that declared core.
- A registry-only suite puts shared harness primitives in a named
supportmodule. Support code may launch fixtures, translate test inputs, poll bounded readiness, or decode outputs. It must not own product behavior, scenario policy, or assertions that determine whether a feature is correct. - New and structurally refactored suites import only the support primitives a behavior module uses. When touching an older suite that exposes a deliberate parent prelude, migrate the affected behavior boundary away from wildcard imports instead of extending the implicit surface.
Refactoring tests
- Split a mixed or oversized test file by behavioral responsibility. Do not compact readable tests, rename an unrelated module, or add a source-limit exemption merely to satisfy a line ceiling.
- Keep setup and assertions beside the behavior they prove when they are used by one module. Promote them to shared support only after a second behavior owner needs the same terminal-independent harness operation.
- Give every behavior module a module-level comment that states its boundary.
Do not create catch-all modules such as
misc,common_tests, orhelpers. - Preserve test names and oracle strength during structural moves unless the behavior contract itself changes.
Evidence
- Test one semantic owner at the narrowest layer that can prove it. Add a cross-layer integration test only when the contract crosses those layers.
- A regression test must fail for the relevant defect, and its oracle must inspect the authoritative state rather than a convenient intermediate acknowledgement.
- Keep failure paths in the same behavior module as the successful contract so their ownership cannot drift.
- Prove key aliases table-first at their typed semantic owner, including plain, Shift, Primary, Primary+Shift, Alt, and mixed modifier reports that the terminal adapter supports. Add integration coverage only for mode ownership, insertion boundaries, durable effects, and configured-binding precedence.
- Terminal input changes require real PTY encodings with durable or editor-state oracles. A successful key send or escape-sequence write is not evidence of behavior.
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.

