bug(mcp): lco_find blocks on implicit full reindex and times out in Hermes

[100yenadmin]

Priority

P0 Hermes-first 1.6 release blocker.

Parent

• Milestone 14: LCO 1.6 Control Plane Release Train
• Parent epic: EPIC 1.6.0 "The Control Plane" — TargetAdapter (two-protocol) + review-then-drive-under-audit marquee + session-diff #673
• Hermes stabilization tracker: TRACKER: LCO 1.6 Hermes-first stabilization and release truth #791

Problem

The MCP lco_find tool treats an omitted index argument as permission to
synchronously index Codex and Claude sources before every query. On Eva's
Hermes-mounted LCO corpus, calls repeatedly reach the configured 300-second MCP
timeout.

The indexed database is already available and queryable. A warm recall request
must not be blocked by an implicit, unbounded cache refresh.

Supported-path impact

• Hermes connection and tool registration succeed.
• lco_find is the primary recall entry point.
• The default call shape used by an agent omits index.
• The default path repeatedly times out at production corpus scale.

No raw query text, thread identifier, transcript content, or customer data is
required to reproduce or prove the latency failure.

Acceptance criteria

• MCP lco_find with omitted index queries the existing index without
  running Codex or Claude indexing.
• index:false remains an explicit no-refresh request.
• index:true retains explicit synchronous refresh behavior.
• lco_index_sessions remains available for explicit refresh.
• Results distinguish index_skipped_by_default from
  index_skipped_by_flag.
• CLI lco find behavior remains unchanged unless its own tests prove it
  shares the MCP defect.
• Twenty warm calls on a protected Eva-sized database snapshot have p95 latency
  at or below 300 ms and maximum latency at or below one second.
• Evidence contains aggregate timing and counts only.

Validation

• Focused MCP/find behavior tests for omitted, false, and true indexing modes.
• Candidate package smoke using the same candidate SHA as the PR.
• Hermes Python MCP client canary against a protected local snapshot.
• Current-head GitHub Actions CI and CodeQL.

Stop conditions

• Stop if meeting the threshold requires returning raw transcript data,
  weakening source-safety filters, or mutating Eva's active profile/database.
• Do not relax the threshold without recording the observed profile and an
  owner decision.

Evidence

/Volumes/LEXAR/Codex/evidence/lco-hermes-first/2026-07-28/

Current status

Verified on published LCO 1.5.0 and still present in source
4384302ecd0caae2a2658a554951a4c3827aa29c.

Exact next action

Implement the MCP-only default change and its focused regressions on
fix/1.6-hermes-stabilization.

[coderabbitai]

🔗 Related PRs

#605 - Implement 1.3.1 day-one UX wave [merged]
#610 - Bound direct CLI recall smoke [merged]
#657 - fix: surface Codex index cap skips [merged]
#697 - feat: add first-run LCO find facade [merged]
#765 - Add LCM prepared cards and peer integrity doctor [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[evaos-code-review-bot]

evaOS issue enrichment

Issue: #790 - bug(mcp): lco_find blocks on implicit full reindex and times out in Hermes
State: open; milestone: Milestone 14: LCO 1.6 Control Plane Release Train

Related issues/PRs: #673, #791.
Existing labels: bug, safety, mcp, orchestrator, eval.
Suggested labels: security, tests, support.
Suggested owners: none.

Related context

• EPIC 1.6.0 "The Control Plane" — TargetAdapter (two-protocol) + review-then-drive-under-audit marquee + session-diff #673 - mentioned in issue metadata; inspect for dependency, duplicate, or prior-decision risk.
• TRACKER: LCO 1.6 Hermes-first stabilization and release truth #791 - mentioned in issue metadata; inspect for dependency, duplicate, or prior-decision risk.

Agent-start packet

Problem shape: bug(mcp): lco_find blocks on implicit full reindex and times out in Hermes - ## Priority P0 Hermes-first 1.6 release blocker. ## Parent - Milestone 14: LCO 1.6 Control Plane Release Train - Parent epic: #673 - Hermes stabilization tracker: #791 ## Problem The MCP lco_find tool treats an
Product fit: Product/market issue; check vision fit, customer value, and whether external precedent can reduce time-to-market.
Implementation wedge: Start with a bounded adapter/contract or dry-run packet before changing live runtime behavior.

Build / borrow / buy scan:

• Build: identify the smallest implementation wedge that proves the issue without broad rollout.
• Borrow: inspect same-repo prior art, repo memory, GitNexus context, and allowlisted cross-repo examples first.
• Buy/use: run capped external OSS/library/API/current-market research with citations, observed dates, and why each source matters.
• Library choice: compare maintenance, API fit, license, bundle/runtime cost, and integration proof.

Candidate sources:

• VISION.md / repo policy / repo-memory.md / GitNexus packet.
• Same-repo issues and PRs referenced by issue metadata.
• Allowlisted cross-repo GitHub examples with direct issue/PR/repo URLs.
• External OSS repositories or library/API docs with freshness dates.
• Current market/product examples, capped and cited, only when they change implementation direction.

Acceptance criteria:

• Preserve and execute the issue's stated acceptance criteria.
• Include source citations for any external OSS/library/current-market claims.
• Keep issue-enrichment comments sticky and idempotent.
• Keep labels, owners, reviewers, and roadmap fields suggestion-only.

Proof plan:

• Focused unit/fixture test for the changed planner behavior.
• Dry-run issue-enrichment evidence packet before live comment posting.
• Redaction check for issue body, citations, and generated markdown.
• Contract or degraded-mode proof before runtime promotion.
• Dependency/license/API-fit note before adding or relying on a package.

Known traps:

• Do not treat related links as proof by themselves; each source needs a reason why it matters.
• Do not scan old issue backlogs or exceed repo-level enrichment throttles.
• External research must stay capped, cited, fresh-dated, and redacted.
• Technical severity alone can miss product/UX regressions; include product-manager review framing.

Non-goals:

• Do not auto-apply labels, owners, reviewers, roadmap fields, or milestones.
• Do not bulk-enrich old backlog issues from this planner output.
• Do not claim external market or OSS research was performed unless cited sources are present.

Context-source taxonomy:

• [enabled] vision_repo_policy_memory_gitnexus: Use repo vision, policy, memory, and GitNexus first before external research.
• [enabled] same_repo_issues_prs: Issue metadata contains same-repo references.
• [enabled] allowlisted_cross_repo_github: Triggered by issue class; keep to configured allowlist and source caps.
• [enabled] external_oss_examples: Use for build/borrow/buy precedent with citations and observed dates.
• [enabled] library_api_docs: Use when implementation may depend on a module, SDK, API, or framework choice.
• [enabled] current_market_examples: Use for product/UX/market positioning issues with freshness dates.
• [enabled] internal_mcp_docs: Use internal docs/MCP context when available; missing context is degraded, not blocking.

Validation suggestions

• Confirm owner, acceptance criteria, and validation evidence before implementation.

Triage gaps

• No obvious issue triage gap detected from issue metadata.

No labels, owners, reviewers, or roadmap fields were changed by this bot.

[100yenadmin]

PR #792 is merge-ready at 5ec7b90c68a25c138d21b3846834a1de376487cb.

Exact-head proof:

• CI/test, CodeQL, and CodeRabbit status passed.
• Independent evaOS semantic review completed; every thread has a recorded terminal disposition and is resolved.
• Candidate package smoke and Hermes smoke/readiness passed with all 14 required Eva tools.
• Hermes-installed Python MCP canary passed on a temporary protected snapshot: initialization/list 50.784 ms; 20 warm searches p95 9.098 ms, max 9.166 ms.
• Package SHA-256: 44f4b562e9a7fa52074d723bc17904739e0067a32cfc4de467583f56a95ae24b.

PR evidence and review dispositions: #792 (comment)

Proof boundary: PR-ready and isolated Hermes-client compatible only. Not merged, published, installed into Eva's active profile, or live-runtime verified. PR #785 remains separate; any merge/release phase must revalidate both heads and declare the merge order.

[100yenadmin]

PR #792 is clean and merge-ready at 9bb2137f69095630b771ea5dac530a6663cbbadc.

Exact-head proof:

• CI/test, CodeQL, and CodeRabbit status passed.
• Independent evaOS semantic review completed; every current review thread has a recorded terminal disposition and is resolved.
• Candidate package smoke and Hermes smoke/readiness passed with all 14 required Eva tools.
• Hermes-installed Python MCP canary passed on a temporary protected snapshot: initialization/list 53.584 ms; 20 warm searches p95 9.197 ms, max 9.226 ms.
• Package SHA-256: 106501e36432581f1de4b94923f096294fb4349b1362db9833f34a24d35323cb.

Final-delta evidence and dispositions: #792 (comment)

Proof boundary: PR-ready and isolated Hermes-client compatible only. Not merged, published, installed into Eva's active profile, or live-runtime verified. PR #785 remains separate; any merge/release phase must revalidate both heads and declare the merge order.

[100yenadmin]

PR #792 is clean and merge-ready at aecd7add3b1e99d52c41ea5ec71059790dc50768.

Exact-head proof:

• CI/test, CodeQL, and CodeRabbit passed.
• Independent evaOS semantic review completed; every current review thread has a recorded terminal disposition and is resolved.
• Candidate package smoke and Hermes smoke/readiness passed with all 14 required Eva tools.
• Hermes-installed Python MCP canary passed on a temporary protected snapshot: initialization/list 58.739 ms; 20 warm searches p95 9.124 ms, max 9.660 ms.
• Package SHA-256: b2c6421ccfbd414ed1944a2713d08309b7f1b5f6b995a6b13e8e7ae16a63b2b5.

Final evidence and dispositions: #792 (comment)

Proof boundary: PR-ready and isolated Hermes-client compatible only. Not merged, published, installed into Eva's active profile, or live-runtime verified. PR #785 remains separate; any merge/release phase must revalidate both heads and declare the merge order.