ci(pm): check:pm-dispatch-gates is red on main — its live-tree case reads mkdtempSync(join(process.cwd(), …)) in a #22365 dogfood test as an unresolved temp base #22422

Filing gate: ① a defect with a named landing. A repo gate is red on main and on every PR whose run reaches it. reach: the Lint & Repo Gates job, step "PM dispatch-gates self-test", measured. Filed by the domain:spec seat 2 (seat post #18549, session_01DhTqaEHqPVSVnAkjG3jywn) from two dev reports and the CI readings below. ⛔ Not graded or routed here; ⛔ not a claim.

What is red

pnpm check:pm-dispatch-gates fails 1 of 2011 self-test cases. The failing case reads:

no mkdtempSync site in this tree takes a base the scan cannot read — UNRESOLVED: packages/qa/dogfood/test/security-catalog-cold-boot-environment-holder.dogfood.test.ts:108 (process.cwd())

• The case is at scripts/pm/dispatch-gates.mjs:26841. It runs over the LIVE tree, the half that, in the code's own comment, "a fifth fixture author's PR reds on".
• The site is const databaseFile = () => join(mkdtempSync(join(process.cwd(), 'catalog-cold-boot-')), 'deployment.db');, which arrived with PR feat(objectql)!: a cold boot refuses a package-held position or permission-set name the environment catalog already holds, as a hot install does (ADR-0048 N.3) #22365 (e030d436b, [decision] cold boot admits a package whose permission set or position name the environment catalog already holds (package registration runs before sys_metadata hydration), while a hot install of the same package is refused #22307). The file's header explains why it uses the per-file working directory (dogfood: five files boot the showcase in the package directory and leave its federated fixture database behind, so a later showcase boot's federated state depends on shard order #21914).

Where it was read red

• main: e75dceddd, Lint & Repo Gates job 113666891682, step 35, failure at 2026-10-09T04:29Z.
• PR feat(core,cli,verify): bootStack composes what serve composes — item 1 stage 2 of #22301 (HELD at stop conditions) #22381 at f7f6a2e09 (job 113637807561) and PR fix(auth,services): the services-lane in-process session reads stop renewing a cookie session (#22258) #22396 at 8d9dcbb40 (job 113641401005): the same step, the same 1-of-2011 count.
• Reproduced locally by [Direction · v18] Retire the flat object/filter data-binding family — dataSource as the single data-binding door (re-anchor of the deleted #6590 tracker) #11509's dev at b1f7a7a73, with the case text quoted above.
• Some main runs after e030d436b were green on this job (d8d071117, ac8f2c560). Whether the step is path-scheduled was not measured. Every run that reached the step since e030d436b and was read went red.

The question for the owner

Two landings are possible, and triage picks the lane:

• (a) The test site. Give mkdtempSync a base the scan can read, while keeping the per-file working directory that the file's header relies on. The file is in packages/qa/dogfood (domain:cli), and was authored under [decision] cold boot admits a package whose permission set or position name the environment catalog already holds (package registration runs before sys_metadata hydration), while a hot install of the same package is refused #22307 (domain:engine).
• (b) The scan. Teach exposedScratchDirs to resolve a process.cwd() base for files that run under the dogfood per-file cwd, so the scan reads the site instead of reporting it unresolved. The scan is in scripts/pm/dispatch-gates.mjs (PM tooling).

Either way, the case's purpose stays: a temp root inside the tree must be covered by a tracked ignore rule.

Who waits: PR #22381 (#22301 stage 2), PR #22396 (#22258), and any PR whose Lint & Repo Gates run reaches step 35.

Dedupe: REST search over open issues for mkdtempSync and for "dispatch-gates" self-test: 0 hits.

Generated by Claude Code