Review Evidence¶
These are point-in-time review artifacts for APTL commit
f35f86ee3bcc41f0b918fd209a944bf91e83f659, captured on 2026-09-05.
Files labelled final record verification after incorporating upstream
65390508db54d118b4dd685db742f3a282b4f366 and the review documents.
| Artifact | Meaning |
|---|---|
| Pre-commit log | All configured baseline hooks passed; quiet hooks do not expose test counts |
| Static gate log | Two static/no-start tests passed; no live range claim |
| Process probe / results | Real harmless subprocess tests showing input deadline overrun and descendant survival |
| Boundary probe / results | Real Python methods against a fake daemon and temporary synthetic env/files |
| Dev protection snapshot | Required status contexts and review configuration read through GitHub API |
| Final pre-commit log | Configured hooks on the updated branch; verbose output records test counts and suite skips |
| Final static gate | 2 passed in 58.30 seconds; still no-start evidence |
| Final process results / boundary results | Same defects reproduced at the updated source baseline; 0.2-second input deadline returned success after 2.022 seconds |
| Upstream changed paths | Explicit delta from the initial tracked-file inventory |
Run the probes from the baseline checkout with its installed development environment:
.venv/bin/python docs/reviews/962-lilrae-readiness/evidence/process-probes.py
.venv/bin/python docs/reviews/962-lilrae-readiness/evidence/boundary-probes.py
The process probe launches local Python children and explicitly kills the synthetic surviving child. The boundary probe uses mocked daemon calls, temporary files and an isolated synthetic environment. It does not launch Docker, read real credentials, or change a user file. Results deliberately describe the baseline defects rather than asserting that they are fixed.
No range start/stop, firewall modification, volume reset, model call, cloud deployment or release was performed. Future qualification must run on a clean supported host/isolated daemon using exact acquired artifacts.
The strict MkDocs build and direct Vale check also passed. The documentation build emitted revision-date plugin warnings, including an existing ADR; these are timestamp metadata warnings rather than broken document links. The Python suite emitted a Starlette/httpx deprecation warning. Neither changes the qualification limits above.
The completed run passed 4,578 Python tests, 462 common MCP tests, 216 red MCP tests and 238 frontend tests. Python reported 94 skips: live smoke tests were not enabled, two cases require Windows semantics, and one supply chain test skips because its fixture was removed or relocated. #969 must reconcile that missing-fixture skip and make release qualification skips explicit.
An earlier final-check attempt passed 4,578 Python tests with 94 skips, but pre-commit stopped because the review documents changed during its run. The recorded final run was repeated after the files settled. Passing tests do not resolve the reproduced defects. Suite selection/skips are visible in the log; live range and provider qualification were not exercised.