fix(monitors): reauthorize before process spawn

[andrei-hasna]

Summary

• recheck the current thread writer fence, permission profile, and cwd after persisting the monitor start event
• use only the refreshed execution snapshot and config for sandboxed process spawn
• preserve the existing generation, routing, sandbox, and cleanup behavior

Regression

• thread_monitor_runtime_does_not_spawn_after_permission_changes_during_start_event_persistence pauses after event persistence, mutates the live permission profile, resumes, and requires failure without a process id
• authoritative AWS RED at 4aef7a22b95bf3a3925137b53175d74fadc96353: named stale-authorization assertion failed
• AWS GREEN and hosted exact-head checks pending

Local gates

• just fmt
• git diff --check and cached diff check
• staged Shield: no security issues
• staged and exact-range Gitleaks: zero findings

Todos: f4a9b60f-b202-4a0d-b7bc-bb6ac2f43099
Focused re-review remains assigned only to Lovelace session 019fe802-b5ae-7033-ba6d-e6f7185dd943 after remote gates.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[andrei-hasna]

[REVIEW] GO — #535 @ 05c98cb — lens: correctness+security+gates, reviewer Damon (1 of 1)

Reviewed exact candidate

• Base: origin/main at c3c84a8.
• git log --oneline origin/main..HEAD: exit 0; 4 commits.
• git diff origin/main...HEAD --stat: exit 0; 2 files, 199 insertions, 6 deletions.
• Full diffs and surrounding source read for codex-rs/app-server/src/request_processors/thread_monitor_runtime.rs and codex-rs/app-server/src/message_processor_schedule_tests.rs, plus the monitor authorization model, command authorization path, and thread-settings update path.

What I ran

• bun install: exit 0; setup only, 521 packages installed. This is not a test gate.
• gh pr view 535 --repo hasna/codewith --json statusCheckRollup: exit 0; final counts: 25 SUCCESS, 0 FAILURE, 3 SKIPPED, 0 pending.
• Initial sdks attempt: 37 passed, 1 timed out, 38 total. gh run rerun 31370486973 --repo hasna/codewith --failed: exit 0; the rerun completed SUCCESS.
• No lane-chosen test command was run. The repository declares no test script and therefore no repository test gate. It also declares no typecheck script.

Forge checks read

• Bazel test on ubuntu-24.04 for x86_64-unknown-linux-gnu — SUCCESS
• cla — SUCCESS
• Check for spelling errors — SUCCESS
• Blob size policy — SUCCESS
• cargo-deny — SUCCESS
• build-test — SUCCESS
• Build Codewith Linux CLI — SUCCESS
• Detect changed areas — SUCCESS
• python-sdk — SUCCESS
• Bazel test on ubuntu-24.04 for x86_64-unknown-linux-musl — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm shard 1/4 — SUCCESS
• Format / etc — SUCCESS
• sdks — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm shard 2/4 — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm shard 3/4 — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm shard 4/4 — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm (native main) — SKIPPED
• cargo shear — SUCCESS
• Argument comment lint package — SKIPPED
• Bazel clippy on ubuntu-24.04 for x86_64-unknown-linux-gnu — SUCCESS
• Bazel clippy on windows-latest for x86_64-pc-windows-gnullvm — SUCCESS
• Verify release build on ubuntu-24.04 for x86_64-unknown-linux-gnu — SUCCESS
• Argument comment lint - Linux — SUCCESS
• Verify release build on windows-latest for x86_64-pc-windows-gnullvm — SUCCESS
• Argument comment lint - Windows — SUCCESS
• Bazel test on windows-latest for x86_64-pc-windows-gnullvm — SUCCESS
• CI results (required) — SUCCESS
• [code]smith — SKIPPED

Blocking P0/P1 findings

• None. The final pre-spawn authorization read binds the launch to the current writer fence, permission profile, thread CWD, and runtime config after asynchronous start-event persistence. The race regression test forces the stale-authorization window and verifies no process is created and the monitor is marked failed.

Non-blocking follow-up

• The first SDK attempt hit the existing 5-second timeout in sdk/typescript/tests/run.test.ts; it has no code-path overlap with this Rust monitor change and passed on rerun. This is flaky-test hardening, not a blocker for this candidate.