You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Say in the property guidance that it covers every target and that fuzzing follows it #15825
/testing:write's property guidance says plainly that it covers every target and that fuzzing is the same mechanism, so nobody builds or looks for a separate fuzzing or UI-property skill.
Demo: ask /testing:write for a property test of a running UI, or for a fuzz test of a parser; it follows the one property guidance, finds the consumer's own tool, and names none itself.
Source: the talk "All Code Is Slop (Except Mine)" (Lauren Tan), slide 28 and the transcript at 1:12:39: "property-based testing as well for UIs". The tool she shows describes itself as "property-based testing, or fuzzing if you will".
Decisions
Owner, 2026-11-12: property testing is one thing with one home and is not split by target ("I think we just consolidate property testing and then split it based on the target"); flood-style fuzzing belongs with it.
A fuzz test is generated input with the weak rule "does not crash"; a property test is generated input with a stated rule. Same mechanism, same guidance.
Generated user actions against a running UI are a generated call sequence on a stateful thing, which the guidance already covers.
No tool is named or shipped; the tool comes from the consumer's repository.
This ships separately from the pilot cut so that cut's before-and-after result is not confounded by added text.
Key interfaces
None.
Done when
The property guidance states, in one or two sentences, that it applies to every target (a running UI included) and that a fuzz test follows it; it names no tool.
One new eval case, frozen first, covers the sentences; the skill's existing write cases are non-inferior on a two-version compare (opus and sonnet aliases, three runs per side, --case filter, a --max-cost-usd ceiling, no new batch after a usage-limit error), run in the one local eval-runner session; the pull request carries do-not-merge until the record lands.
The pull request body names this repository's dogfooding step.
Size: Small
Complexity: Low
Uncertainty: Known
Flags: none
Blocked by: none
Understood: add one or two sentences to the /testing:write property guidance saying it applies to every target (a running UI included) and that a fuzz test follows it. The sentences name no tool. The change also needs one new frozen eval case and a non-inferiority compare on the existing write cases, and the pull request body names the dogfooding step.
Not admitted to the issue lane: merged pull request #16464 names it with Refs: after its newest decision or refinement comment, so its work may already be done: close the issue if it is, or post a comment whose first line is **Decision recorded.** saying what is left, which sends it back through intake. Admission takes the issue once each of those is met, when intake triages it again (.github/actions/resolve-config/admission-rules.json).
melodic-automation-lanes commented on Nov 15, 2026
Pull request #16464 merged with Refs:, so this issue stays open and nothing starts the lane on it again. work-unblock removed status: ready and agent-ready, so the hourly triage sweep sends it back to intake triage. Admission then parks it at status: confirmed for a check of what that pull request left, until a **Decision recorded.** comment says what is left or someone closes it.
Activity
github-actions commentedon Nov 13, 2026
Size: Small
Complexity: Low
Uncertainty: Known
Flags: none
Blocked by: none
Understood: add one or two sentences to the
/testing:writeproperty guidance saying it applies to every target (a running UI included) and that a fuzz test follows it. The sentences name no tool. The change also needs one new frozen eval case and a non-inferiority compare on the existing write cases, and the pull request body names the dogfooding step.Refs:and left this issue open. I could not tell from the issue what part of the done-when it covered. A maintainer should check that before work starts.Automated triage. A maintainer reviews it.
melodic-automation-lanes commentedon Nov 13, 2026
Not admitted to the issue lane: merged pull request #16464 names it with
Refs:after its newest decision or refinement comment, so its work may already be done: close the issue if it is, or post a comment whose first line is**Decision recorded.**saying what is left, which sends it back through intake. Admission takes the issue once each of those is met, when intake triages it again (.github/actions/resolve-config/admission-rules.json).melodic-automation-lanes commentedon Nov 15, 2026
{"v":1,"actor_id":337595936,"title_sha256":"a522e6a34f2e287f8ac73c60603a525bca8dfaa52aa4c35d400932f270c43b36","body_sha256":"2e81b8949b5679bf96df16984207b862439f7aad32d4f2058e3bb8b2fc874ea9"}
github-actions commentedon Nov 15, 2026
melodic-automation-lanes commentedon Nov 15, 2026
Pull request #16464 merged with
Refs:, so this issue stays open and nothing starts the lane on it again. work-unblock removedstatus: readyandagent-ready, so the hourly triage sweep sends it back to intake triage. Admission then parks it atstatus: confirmedfor a check of what that pull request left, until a**Decision recorded.**comment says what is left or someone closes it.github-actions commentedon Nov 16, 2026
The triage sweep dispatched intake triage for this issue (dispatch 1; 2 runs that reach intake, or 6 dispatches, hand it to a maintainer).