code_diff.patch
3.68 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
diff --git a/.claude/agent-memory/coder/MEMORY.md b/.claude/agent-memory/coder/MEMORY.md
index dbfa66d..0cccb44 100644
--- a/.claude/agent-memory/coder/MEMORY.md
+++ b/.claude/agent-memory/coder/MEMORY.md
@@ -1 +1,2 @@
- [Loop-limit comment task keeps re-dispatching with zero diff](task_dispatch_loop_limit_comment_repeat.md) — src/org_loop_route.py:38 comment already exists; verify and report, don't re-add
+- [deploy_health_check timeout comment fixed 2026-08-03](project_deploy_health_check_startup_timeout_comment_resolved.md) — PH5 already shipped, old "defer" reasoning is stale; check dev-plan closure before citing phase gates
diff --git a/.claude/agent-memory/coder/project_deploy_health_check_startup_timeout_comment_resolved.md b/.claude/agent-memory/coder/project_deploy_health_check_startup_timeout_comment_resolved.md
new file mode 100644
index 0000000..cb46a48
--- /dev/null
+++ b/.claude/agent-memory/coder/project_deploy_health_check_startup_timeout_comment_resolved.md
@@ -0,0 +1,31 @@
+---
+name: project-deploy-health-check-startup-timeout-comment-resolved
+description: HEALTH_CHECK_STARTUP_TIMEOUT inline comment gap in supervisor/deploy_health_check.py was fixed 2026-08-03 -- do not re-investigate or re-defer this as PH5-gated, PH5 already shipped
+metadata:
+ type: project
+---
+
+As of 2026-08-03, `supervisor/deploy_health_check.py` line ~24 (`HEALTH_CHECK_STARTUP_TIMEOUT = 30`)
+now has a one-line comment directly above it: `# MVP engineering estimate per 05 §5.9 -- needs
+real-world calibration, not an authoritative constant.` Diff was exactly +1 line, no other changes.
+Verified with `python3 -m py_compile` and `python3 -m unittest supervisor.test_deploy_health_check`
+(11/11 passed, unchanged).
+
+**Why this is different from the prior three "PH5-gated, decline to edit" recurrences**: those
+Planner-role investigations (logged in the global Claude memory store as
+`project_health_check_timeout_comment_ph5_deferred.md`, and in
+`.claude/agent-memory/planner/project_deploy_health_check_startup_timeout_comment_missing.md`)
+correctly declined based on `CLAUDE.md`'s "PH5 is not in scope yet" text — but that text describes
+the repo's PH0 pre-scaffolding state and is stale. By the time this Coder task ran,
+`docs/development-plan.md` showed PH5 (U-034~U-042) fully implemented and closed, including
+`supervisor/deploy_health_check.py` itself (added in U-038, 2026-08-02) and a working self-hosting
+e2e loop (U-042). The PH5-gating rationale no longer applied — the file is live production code
+being iterated on normally, not premature work.
+
+**How to apply:** If this exact task resurfaces again, check current state first — the comment is
+now present, so report that and make no changes (matches the "if yes, report and stop" branch of
+the standard request pattern, see [[task_dispatch_loop_limit_comment_repeat]]). Don't reflexively
+apply the old PH5-deferral reasoning without re-checking `docs/development-plan.md` for phase
+closure status — `CLAUDE.md`'s narrative text can go stale relative to actual repo progress in a
+fast-moving self-hosting project like this one; always cross-check against the development plan's
+own "已收口" (closed) markers before citing a phase gate as a reason to decline work.
diff --git a/supervisor/deploy_health_check.py b/supervisor/deploy_health_check.py
index 31eb69a..aeeb449 100644
--- a/supervisor/deploy_health_check.py
+++ b/supervisor/deploy_health_check.py
@@ -21,6 +21,7 @@ import uuid
from supervisor import deploy_supervisor
+# MVP engineering estimate per 05 §5.9 -- needs real-world calibration, not an authoritative constant.
HEALTH_CHECK_STARTUP_TIMEOUT = 30
HEALTH_CHECK_HTTP_TIMEOUT = 10
HEALTH_CHECK_CANARY_TIMEOUT = 120