fix: add isInvalidAgentIdFormat guard to runMemoryReflection (Issue #686)

[jlin53882]

Fix: Add isInvalidAgentIdFormat guard to runMemoryReflection (Issue #686)

Summary

Adds a missing isInvalidAgentIdFormat() guard to runMemoryReflection, making it consistent with the other 5 hook sites that all have this check.

Changes

index.ts — runMemoryReflection command hook:

  const sourceAgentId = parseAgentIdFromSessionKey(sessionKey) || "main";
+ // Guard: skip if agentId is invalid format (consistent with other hook sites)
+ if (isInvalidAgentIdFormat(sourceAgentId, config.declaredAgents)) {
+   api.logger.debug?.(
+     `memory-reflection: command hook skipped \u2014 invalid agentId '${sourceAgentId}'`,
+   );
+   return;
+ }
  // Exclude agents/sessions listed in memoryReflection.excludeAgents (supports wildcards)

Why this matters

All other hook sites in PR #516 have isInvalidAgentIdFormat() as a 前置 guard:

• before_prompt_build (auto-recall) — ✅ has guard
• before_prompt_build (recallWork) — ✅ has guard
• before_prompt_build (auto-capture) — ✅ has guard
• before_prompt_build (reflection inheritance) — ✅ has guard
• before_prompt_build (reflection derived+error) — ✅ has guard
• before_reset — ✅ has guard
• runMemoryReflection — ✅ now has guard (this PR)

Currently protected by the || "main" fallback, but the explicit guard is the correct defense-in-depth pattern.

Dependencies

This PR builds on #516. It should be reviewed and merged after #516.

Links

• Issue: runMemoryReflection missing isInvalidAgentIdFormat guard (inconsistent with other hook sites) #686
• Parent PR: fix: complete Issue #492 protection -- per-agent exclusion + internal session guards #516

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a1b3bd46f2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Allow zero serial cooldown to disable reflection throttle

memoryReflection.serialCooldownMs is documented as “set to 0 to disable,” but this parser uses parsePositiveInt, which rejects 0 and falls back to the default 120000 ms. In configs that explicitly set serialCooldownMs: 0, reflection still remains throttled for two minutes, so command-hook reflections are unexpectedly skipped instead of running every trigger.

Useful? React with 👍 / 👎.

[jlin53882]

CI 失敗分析（3 個 Job 均無關聯 PR #687）

失敗摘要

Job	錯誤檔案	根因	是否關聯 PR #687
core-regression	test/smart-extractor-branches.mjs:497	測試斷言失敗	❌ 檔案無變動
storage-and-schema	smart-extractor.ts:444 - this.store.bulkStore is not a function	mock 缺少 bulkStore	❌ 測試 mock 既有問題
packaging-and-workflow	scripts/verify-ci-test-manifest.mjs:60	manifest 未列入 test/import-markdown/	❌ CI 設定既有問題

關鍵觀察

三個 Job 的 Test step 都失敗在同一個 merge commit (011d344841633aa) 上，這是 PR #687 的 parent commit（merge base），並非 PR 自己引入的變更。PR #687 只修改了 index.ts，與上述三個測試檔案完全無交集。

建議

建議開獨立 issue 追蹤這三個 upstream 既有問題，或在 upstream/master 上驗證這些失敗是否已存在後再合併 PR #687。