chore: git cache cleanup

docs/issues/manual_test_pr1_caddy_compatibility_closure.md

@@ -0,0 +1,95 @@
+## Manual Test Tracking Plan — PR-1 Caddy Compatibility Closure
+
+- Date: 2026-02-23
+- Scope: PR-1 only
+- Goal: Track potential bugs in the completed PR-1 slice and confirm safe promotion.
+
+## In Scope Features
+
+1. Compatibility matrix execution and pass/fail outcomes
+2. Release guard behavior (promotion gate)
+3. Candidate build path behavior (`CADDY_USE_CANDIDATE=1`)
+4. Non-drift defaults (`CADDY_USE_CANDIDATE=0` remains default)
+
+## Out of Scope
+
+- PR-2 and later slices
+- Unrelated frontend feature behavior
+- Historical QA items not tied to PR-1
+
+## Environment Checklist
+
+- [ ] Local repository is up to date with PR-1 changes
+- [ ] Docker build completes successfully
+- [ ] Test output directory is clean or isolated for this run
+
+## Test Cases
+
+### TC-001 — Compatibility Matrix Completes
+
+- Area: Compatibility matrix
+- Risk: False PASS due to partial artifacts or mixed output paths
+- Steps:
+  1. Run the matrix script with an isolated output directory.
+  2. Verify all expected rows are present for scenarios A/B/C and amd64/arm64.
+  3. Confirm each row has explicit PASS/FAIL values for required checks.
+- Expected:
+  - Matrix completes without missing rows.
+  - Row statuses are deterministic and readable.
+- Status: [ ] Not run [ ] Pass [ ] Fail
+- Notes:
+
+### TC-002 — Promotion Gate Enforces Scenario A Only
+
+- Area: Release guard
+- Risk: Incorrect gate logic blocks or allows promotion unexpectedly
+- Steps:
+  1. Review matrix results for scenario A on amd64 and arm64.
+  2. Confirm promotion decision uses scenario A on both architectures.
+  3. Confirm scenario B/C are evidence-only and do not flip the promotion verdict.
+- Expected:
+  - Promotion gate follows PR-1 rule exactly.
+- Status: [ ] Not run [ ] Pass [ ] Fail
+- Notes:
+
+### TC-003 — Candidate Build Path Is Opt-In
+
+- Area: Candidate build path
+- Risk: Candidate path becomes active without explicit opt-in
+- Steps:
+  1. Build with default arguments.
+  2. Confirm runtime behavior is standard (non-candidate path).
+  3. Build again with candidate opt-in enabled.
+  4. Confirm candidate path is only active in the opt-in build.
+- Expected:
+  - Candidate behavior appears only when explicitly enabled.
+- Status: [ ] Not run [ ] Pass [ ] Fail
+- Notes:
+
+### TC-004 — Default Runtime Behavior Does Not Drift
+
+- Area: Non-drift defaults
+- Risk: Silent default drift after PR-1 merge
+- Steps:
+  1. Verify Docker defaults used by standard build.
+  2. Run a standard deployment path.
+  3. Confirm behavior matches pre-PR-1 default expectations.
+- Expected:
+  - Default runtime remains non-candidate.
+- Status: [ ] Not run [ ] Pass [ ] Fail
+- Notes:
+
+## Defect Log
+
+Use this section for any issue found during manual testing.
+
+| ID | Test Case | Severity | Summary | Reproducible | Status |
+| --- | --- | --- | --- | --- | --- |
+| | | | | | |
+
+## Exit Criteria
+
+- [ ] All four PR-1 test cases executed
+- [ ] No unresolved critical defects
+- [ ] Promotion decision is traceable to matrix evidence
+- [ ] Any failures documented with clear next action