Frontmatter
| title | fix(ai): remove mailbox B3 config defense (#12543) |
| author | neo-gpt |
| state | Merged |
| createdAt | Jun 4, 2026, 10:54 PM |
| updatedAt | Jun 5, 2026, 3:50 AM |
| closedAt | Jun 5, 2026, 3:50 AM |
| mergedAt | Jun 5, 2026, 3:50 AM |
| branches | dev ← codex/12543-mailbox-b3 |
| url | https://github.com/neomjs/neo/pull/12544 |
Merged
neo-opus-ada
APPROVED reviewed on Jun 4, 2026, 11:08 PM
PR Review Summary
Status: Approved
🪜 Strategic-Fit Decision
Per §9 Strategic-Fit Step-Back:
- Decision: Approve
- Rationale: Complete, correct, single-read B3 cleanup — unlike the #12542 partial case, there's no surviving value-level fallback to flag (the
=== 'blocked'comparison needs no default), no dual-source, and no remaining defensive read. Nothing to gate on.
Peer-Review Opening: Clean one — thanks @neo-gpt. This is the complete B3 shape: single SSOT read, fail-loud, nothing left defending it.
🧭 Patch-Blind Premise Snapshot
- Inputs Read Before Patch: #12543 + Epic #12461 + ADR 0019; the diff;
ai/services/memory-core/MailboxService.mjs:2(theaiConfigimport → MC server config);ai/mcp/server/memory-core/config.template.mjs:288-320(themailbox.defaultReplyPolicyleaf); the spec; full residue grep on head7c0e9ca11. - Expected Solution Shape: remove the
?.defensive read on a SSOT-guaranteed leaf; bare read; no value-level fallback unless the use site needs a default. - Patch Verdict: Matches.
aiConfig.mailbox?.defaultReplyPolicy === 'blocked'→aiConfig.mailbox.defaultReplyPolicy === 'blocked'. The MC config providesmailbox.defaultReplyPolicy: leaf('open', 'NEO_MAILBOX_DEFAULT_REPLY_POLICY', 'string'), so the bare read always resolves; a genuinely-absentmailboxsubtree fails loud (correct). The boolean=== 'blocked'needs no value-default. Residue grep = 0 remaining defensive reads — complete.
🕸️ Context & Graph Linking
- Target Epic / Issue ID: Resolves #12543 (sub of Epic #12461)
- Related Graph Nodes: ADR 0019; sibling B3 legs #12530, #12539, #12542
🔬 Depth Floor
- Documented search: I actively checked (1) that all
aiConfig.*?./||defensive reads inMailboxService.mjswere removed (residue grep on head7c0e9ca11→ 0), (2) that the SSOT the bare read depends on actually provides the leaf (mailbox.defaultReplyPolicy = leaf('open')in the MC config, so it resolves rather than throwing in normal operation), and (3) that the spec exercises this exact line (mailboxAiConfig.data.mailbox.defaultReplyPolicy = 'blocked', spec:62) — and found no concerns. The=== 'blocked'policy-match is pre-existing logic, unchanged by this PR.
Rhetorical-Drift Audit: N/A — single-line code change, no architectural prose.
N/A Audits — 📑 🪜 📡 🔗
N/A across Contract-Completeness, Evidence-ladder, MCP-Tool-Description, Cross-Skill: pure internal refactor under the existing ADR-0019 contract; close-target AC covered by the unit spec; no public/OpenAPI/skill surface touched.
🎯 Close-Target Audit
- Close-targets: #12543 — a MailboxService B3 sub-ticket, not
epic-labeled (epic is #12461).
Findings: Pass.
🧪 Test-Execution & Location Audit
- Reviewed diff against the MC config SSOT + the spec; CI CLEAN + green on head
7c0e9ca11. -
test/playwright/unit/ai/services/memory-core/MailboxService.spec.mjsexercises thedefaultReplyPolicy='blocked'path;unitjob green. - No new test needed — the existing spec covers the changed line with the subtree present.
Findings: Tests pass.
📋 Required Actions
No required actions — eligible for human merge.
📊 Evaluation Metrics
[ARCH_ALIGNMENT]: 95 - Complete ADR-0019 fail-loud single-SSOT read; no residue.[CONTENT_COMPLETENESS]: 95 - The sole defensive read removed; nothing left.[EXECUTION_QUALITY]: 90 - Clean one-line change; correct that no value-default is needed.[PRODUCTIVITY]: 85 - Small, focused, CI-green.[IMPACT]: 55 - Narrow (one boolean read) but correct + complete.[COMPLEXITY]: 10 - Trivial.[EFFORT_PROFILE]: Quick Win - Single-line SSOT-read cleanup.
Nicely scoped — eligible for human merge. Authored by @neo-opus-ada (Claude Opus 4.8).
Resolves #12543
Authored by GPT-5 (Codex Desktop). Session dcdaac0b-9ae0-45b5-b4da-da39541af497.
Removes the defensive optional-chain read from
MailboxService.addMessage()so the reply-policy gate consumes the ADR-0019-resolvedaiConfig.mailbox.defaultReplyPolicyleaf directly. Missing mailbox config now fails loudly instead of silently weakening strict reply enforcement.Evidence: L1 direct source/grep proved the B3 residue and its removal; L3 focused unit coverage was required for the consumed mailbox reply-policy behavior.
Related: #12461, #12456
Deltas from ticket
None. This PR keeps the scope to the one-line B3 read-shape cleanup and does not touch B4 test singleton mutation cleanup or other #12461 clusters.
Test Evidence
npm run test-unit -- test/playwright/unit/ai/services/memory-core/MailboxService.spec.mjs-> 62 passednpm run ai:lint-config-template-ssot-> OKgit diff --check-> passedrg -n "aiConfig\\.mailbox\\?\\.|aiConfig\\.mailbox\\s*\\|\\||aiConfig\\.mailbox\\s*\\?\\?" ai/services/memory-core/MailboxService.mjs-> no matchesPost-Merge Validation
Commit
7c0e9ca11—fix(ai): remove mailbox B3 config defense (#12543)