Skip to content
This repository was archived by the owner on Jun 12, 2026. It is now read-only.

qa-results.md

Latest commit

 

History

History
35 lines (28 loc) · 2.92 KB

qa-results.md

File metadata and controls

35 lines (28 loc) · 2.92 KB
Raw

QA Results

Status: PASSED

Command-backed evidence from May 17, 2026:

  • npm test: passed, 21 files / 65 tests.
  • npm run check: passed, tsgo --noEmit and biome check ..
  • bun --install=fallback /Users/yeongyu/.config/opencode/skills/typescript-programmer/scripts/check-no-excuse-rules.ts src test scripts: passed, no no-excuse violations across 53 files.
  • LSP diagnostics: passed, 49 TypeScript files scanned, 0 diagnostics.
  • npm run qa:import: passed, import ok.
  • npm run qa:status-scope: passed, status scope ok.
  • npm run qa:child-tools: passed, child tools ok.
  • npm run qa:process-kill: passed, child pid was reported and direct SIGTERM produced killed.
  • npm run qa:senpi-install: passed dry-run for ~/.senpi/agent/extensions/pi-task.
  • node scripts/qa-senpi-install.mjs --force: passed, local symlink installed to ~/.senpi/agent/extensions/pi-task.
  • readlink ~/.senpi/agent/extensions/pi-task: passed, points to /Users/yeongyu/local-workspaces/pi-extensions/pi-task.
  • senpi --help: passed, host exposes --extension, --no-session, --tools, --no-tools, and --resume.
  • npm pack --dry-run: passed, pi-task-0.1.4.tgz, 43 files.

Manual QA scenario note:

  • /var/folders/nj/hqfr8ndn5q56cqw7jqgbrck40000gn/T/ulw-scenarios.XXXXXX.md.WidgZz2Lwo
  • Direct TUI scoping simulation verified that a fresh/current session footer sees only that session's tasks and does not leak another session's writer task.
  • Direct child-policy simulation verified tools: rules resolve to ["read","task","task_cancel","task_status"] and in-process child creation receives persistSession: false.
  • Read-only slop/type reviewer initially failed two issues: code-defined agent tools were not normalized into permissions, and direct process-runner deny-only handling used a hardcoded inherited allowlist. Both were fixed and covered by regression tests.
  • GPT-5.2 xhigh verifier initially blocked five plan-completeness gaps: missing manager scoped list API, missing parent/root/model-attempt/exit facts in UI, depth-unaware task tool enabling, dropped process exit metadata, and missing live task-change UI update path. All five were fixed and covered by regression tests.
  • Read-only slop/type re-review 019e3056-a318-76c1-9f79-9c913b8a7cba: PASS after the cancellation-signal blocker was fixed.
  • GPT-5.2 xhigh final verifier 019e3056-a366-7fe3-9ac6-f38d0af930d1: PASS for plan/user-requirement conformance.

Residual notes:

  • pi is not installed on this machine; senpi is installed and is the verified host command.
  • Full interactive mouse QA depends on the installed pi TUI selector behavior. The extension uses ctx.ui.select() and ctx.ui.confirm() so keyboard support is covered by public API and mouse support follows the host TUI.
  • task_cancel with an external host abort signal is covered by regression test and external re-review: the runner observes the same internal cancellation signal that TUI cancel aborts.