Greptile Summary

This PR fixes a real bug where the health endpoint returned running: false, connected: false for channels while channels.status returned correct values. The fix threads the channel manager's getRuntimeSnapshot through startGatewayMaintenanceTimers → setRuntimeSnapshotFn → refreshGatewayHealthSnapshot → getHealthSnapshot, where runtime fields are spread before the overriding accountId/enabled/configured fields.

Core correctness: The merge order (...runtimeSnapshot, accountId, enabled, configured) is correct; authoritative health-check fields take precedence over runtime state. The implementation properly handles edge cases: CLI health command (runtime param is optional), channels without runtime entries (no spread applied). The spread operator safely merges runtime state without overwriting intended fields.

What changed:

• getHealthSnapshot() now accepts optional runtime state and merges it into channel snapshots before buildChannelSummary
• refreshGatewayHealthSnapshot calls runtimeSnapshotFn?.() to obtain runtime state and passes it through
• setRuntimeSnapshotFn setter provides a way for startGatewayMaintenanceTimers to inject the runtime snapshot function
• Gateway wires getRuntimeSnapshot from the channel manager through the maintenance timer setup

Confidence Score: 4/5

• Safe to merge — core fix is correct and properly scoped; runtime state threading is secure and backward compatible.
• The PR directly addresses the documented bug (health endpoint returning incorrect channel state). The fix properly merges runtime state via spread operator with field-override precedence to maintain intended behavior. Code handles edge cases correctly: optional runtime for CLI context, safe handling of missing runtime entries, consistent health-check field priority. No issues prevent merging.
• No files require special attention.

_{Last reviewed commit: dc16f0a}

This pull request has been automatically marked as stale due to inactivity.
Please add updates or it will be closed.

Codex automated review: keeping this open.

Keep this PR open. Current main still lacks the runtime-state merge that this PR is trying to add, and the canonical maintainer repair remains open/unmerged with explicit maintainer guidance not to close duplicate or superseded PRs yet.

Best possible solution:

Keep this PR open while maintainer PR #72417 remains open and explicitly asks not to close duplicate or superseded PRs yet. The maintainer path is to review and land #72417 once clean and validated; otherwise, salvage this PR by rebasing it onto current main, threading runtime through the current gateway health refresh path, projecting only safe runtime fields, preserving probe behavior deliberately, and adding regression tests for health snapshots and gateway health refresh.

What I checked:

• current_health_snapshot_has_no_runtime_input: On current main, getHealthSnapshot only accepts timeoutMs and probe; the channel account snapshot passed into plugin summary builders is built from config/probe fields and does not accept or merge channel runtime state. (src/commands/health.ts:363, 9bc703213bb3)
• gateway_health_refresh_does_not_pass_runtime: The gateway health refresh path still calls getHealthSnapshot({ probe: opts?.probe }) directly, caches/broadcasts that result, and has no runtime callback or runtime snapshot projection. (src/gateway/server/health-state.ts:72, 9bc703213bb3)
• channels_status_uses_runtime_state: The channels.status method reads context.getRuntimeSnapshot(), selects channel/account runtime entries, and passes runtime into buildChannelAccountSnapshot, confirming that status and health use different state sources today. (src/gateway/server-methods/channels.ts:161, 9bc703213bb3)
• whatsapp_defaults_absent_runtime_to_false: WhatsApp's health summary builder reports running and connected as false when those fields are absent from the snapshot, so a live runtime state omitted from health can still produce the false-negative symptom described by the linked bug. (extensions/whatsapp/src/channel.ts:243, 9bc703213bb3)
• canonical_maintainer_repair_open: Related maintainer PR fix(gateway): preserve runtime-backed health state #72417 is open, author-associated MEMBER, labeled maintainer, credits this PR's runtime snapshot plumbing approach, and explicitly says not to close duplicate or superseded PRs until that repair is pushed, review-clean, and validated. (2397a253a9ec)
• pr_security_review_no_new_execution_surface_but_data_projection_risk: The PR diff only threads runtime state through health snapshot generation and gateway maintenance. It does not add dependencies, workflows, lifecycle scripts, downloaded artifacts, network calls, or secret-handling changes. The main review risk is functional/data exposure: it spreads a full ChannelAccountSnapshot into health summaries instead of projecting only safe runtime fields. (src/commands/health.ts:454, 2b5583255d02)

Remaining risk / open question:

• This PR is stale against the current gateway startup path and likely needs a rebase before it can merge cleanly.
• The current patch has no regression coverage for runtime-backed getHealthSnapshot or gateway health refresh behavior.
• The PR spreads full runtime channel snapshots into health-summary input; a final fix should prefer a narrow, safe runtime projection so secret-bearing or unrelated snapshot fields cannot leak through plugin summaries.

Codex Review notes: model gpt-5.5, reasoning high; reviewed against 9bc703213bb3.

ProjectClownfish landed the canonical replacement in #72417, merged at be6263d.

Closing this as superseded by that merged fix. Your contribution was carried forward and credited in the changelog/source PR notes.