feat(hotpath): fill daemon mcp/host-admission/io gaps

[ScriptedAlchemy]

feat(hotpath): fill daemon mcp/host-admission/io gaps

Gap-fill hotpath coverage PR for the root daemon crate (crates/tracedecay), branched from codex/tracedecay-total-redesign-plan-reopened (PR #707 base; #707 is NOT merged by this PR).

Rebase note: the base later extracted src/daemon/store_runtime/session_registry.rs into the new tracedecay-store-runtime crate. The 12 session-registry sites in this PR followed that file move, so the diff now touches crates/tracedecay-store-runtime/src/session_registry.rs (same functions, same labels — no new scope). All other files remain in crates/tracedecay.

Scope guarantees:

• Only the 12 originally targeted files touched (11 in crates/tracedecay, plus the rename-followed session_registry.rs above) — no Cargo.toml changes, no hooks/host-integration files, no *_benchmark.rs / *_tests.rs.
• No existing #[hotpath::measure] site was modified, relabeled, or removed.
• hotpath/hotpath-alloc/hotpath-cpu/hotpath-mcp remain out of default and production; macros no-op without --features hotpath. No metrics server, no output-format changes — the MCP hook-protocol stdout guard is untouched.
• All new labels follow daemon.<area>.<verb> / mcp.<area>.<verb> and are unique repo-wide (re-verified after each rebase).

Verification: cargo check -p tracedecay (default features) passes on the rebased branch. cargo check -p tracedecay --features hotpath is currently blocked by a pre-existing base break unrelated to this PR: tracedecay-session-runtime fails with queries overflow the depth limit! at session_sync.rs:431 (base-side instrumentation; needs #![recursion_limit = "256"] or restructuring on the base). This PR's hotpath-featured build passed before that break landed.

Coverage table

102 new measure sites across 12 files (previously zero-coverage files first, then add-missing-only files):

File	New sites	New labels
crates/tracedecay/src/host_admission.rs	14	daemon.host_admission.{install_background_cpu, mount_sibling_project, open_runtime, scan_storage_bytes, digest_session_domain, replay_observations, count_temporal_rows, upsert_transcript_batch, count_transcript_rows, init_project_graph, open_project_graph, open_project_branch, open_graph_read_only, validate_authorities}
crates/tracedecay/src/mcp/server.rs	8	mcp.server.{construct, read_tokens_baseline, detect_worktree_mismatch, reconcile_automation, retrieval_scope_check, mount_work_evidence, mount_retained_surfaces, stats_snapshot}
crates/tracedecay/src/mcp/project_route.rs	9	mcp.project_route.{resolve_registered, forget_project, observe_route, select_route, snapshot_cache, store_cache, refresh_cache, forget_project_shared, protect_structural_ids}
crates/tracedecay/src/mcp/server/ledger.rs	7	mcp.ledger.{read_upload_policy, estimate_raw_tokens, persist_token_accounting, flush_worldwide, record_error_analytics, record_route_analytics, record_span_observation}
crates/tracedecay/src/tracedecay/queries/meta.rs	7	daemon.store_meta.{read_tokens_saved, write_tokens_saved, read_local_counter, reset_local_counter, add_local_counter, checkpoint, quick_check}
crates/tracedecay/src/profile_registry_maintenance.rs	7	daemon.profile_registry.{open_existing, open, list_projects, classify_storage, retire_paths, apply_orphan_relink, gc}
crates/tracedecay-store-runtime/src/session_registry.rs (moved from crates/tracedecay/src/daemon/store_runtime/ by base refactor)	12	daemon.session_registry.{list_ready_sessions, reserve_session_replacement, reserve_session_recovery, reconstruct_terminal_recovery, bind_memory_graph, admit_remote_node, admit_project_runtime, extend_project_runtime, reserve_runtime_retirement, retire_session_sync, rebind_session_sync, runtime_incarnation}
crates/tracedecay/src/daemon/http_application.rs	8	daemon.http.application.{build_registry_router, remote_status_read, local_admission, tls_bind, tls_accept, tls_validate_identity, observe_http_request, parse_body_length}
crates/tracedecay/src/daemon/engine.rs	4	daemon.engine.{git_watcher_health, log_version_skew, claim_catalog_refresh, cached_open_failure}
crates/tracedecay/src/daemon/scheduler.rs	9	daemon.scheduler.{settle_retained_automation, commit_exit, retire_scheduler, retained_project_graph, global_retention, read_automation_config, probe_scheduler_work, user_jobs_pass, mint_user_job_run_id}
crates/tracedecay/src/daemon/maintenance.rs	7	daemon.maintenance.{read_table_growth, read_store_size, compare_table_growth, sample_store_telemetry, mint_telemetry_context, load_cursor, persist_cursor}
crates/tracedecay/src/daemon/pr_autotrack.rs	10	daemon.pr_autotrack.{run_git, discover_gh, discover_ls_remote, activate_manual_branch, cleanup_manual_activation, cleanup_manual_retirement, activate_worktree, remove_store, prepare_worktree, cleanup_worktree}

Macro forms used: #[hotpath::measure(label = ...)] (sync fns), #[hotpath::measure(label = ..., future = true)] (async fns), hotpath::future! for two awaited startup stages inside McpServer::new_with_context and the two Box::pin(async move { ... }) boxed-future bodies in daemon/maintenance.rs (same pattern already used in daemon/engine.rs). The larger diff stat in maintenance.rs is indentation-only from those two wraps.

  

[changeset-bot]

⚠️ No Changeset found

Latest commit: c36a294

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[github-actions]

Performance Comparison codex/tracedecay-total-redesign-plan-reopened → cursor/hotpath-gap-fill-leftovers-04d2

Total Elapsed Time: 5.32s → 5.14s (-3.4%)
CPU Baseline: 80.70µs → 80.08µs (-0.8%)
Benchmark ID: index-bench-timing

timing - Execution duration of functions.

+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| Function                                 | Calls                      | Avg                             | P95                             | Total                            | % Total                      |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| tracedecay-index-bench                   | 1 → 1 (+0.0%)              | 5.32s → 5.14s (-3.4%)           | 5.32s → 5.14s (-3.4%)           | 5.32s → 5.14s (-3.4%)            | 100.00% → 100.00% (+0.0%)    |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.append_pages              | 13 → 13 (+0.0%)            | 174.11ms → 184.48ms (+6.0%)     | 237.24ms → 256.90ms (+8.3%)     | 2.26s → 2.40s (+6.2%)            | 42.57% → 46.69% (+9.7%)      |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.batch.sqlite              | 13 → 13 (+0.0%)            | 118.99ms → 122.62ms (+3.1%)     | 163.05ms → 168.43ms (+3.3%)     | 1.55s → 1.59s (+2.6%)            | 29.10% → 31.03% (+6.6%)      |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.batch.postings            | 13 → 13 (+0.0%)            | 104.30ms → 107.47ms (+3.0%)     | 143.79ms → 149.95ms (+4.3%)     | 1.36s → 1.40s (+2.9%)            | 25.51% → 27.20% (+6.6%)      |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| code_index.workers.install               | 83 → 83 (+0.0%)            | 19.37ms → 16.77ms (-13.4%)      | 56.59ms → 61.34ms (+8.4%)       | 1.61s → 1.39s (-13.7%)           | 30.23% → 27.10% (-10.4%)     |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| CodeIndexProductionOwnerV1::extract_file | 276 → 276 (+0.0%)          | 6.36ms → 4.81ms (-24.4%) 🚀     | 12.59ms → 7.42ms (-41.1%) 🚀    | 1.75s → 1.33s (-24.0%) 🚀        | 33.00% → 25.83% (-21.7%) 🚀  |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| code_index.build.and_publish             | 2 → 2 (+0.0%)              | 634.37ms → 520.08ms (-18.0%)    | 652.21ms → 522.98ms (-19.8%)    | 1.27s → 1.04s (-18.1%)           | 23.86% → 20.25% (-15.1%)     |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| domain.canonical.sha256                  | 114526 → 114526 (+0.0%)    | 10.89µs → 9.04µs (-17.0%)       | 14.92µs → 14.95µs (+0.2%)       | 1.25s → 1.03s (-17.6%)           | 23.47% → 20.15% (-14.1%)     |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.finalization.advance_wake | 14 → 14 (+0.0%)            | 53.15ms → 54.89ms (+3.3%)       | 260.83ms → 275.25ms (+5.5%)     | 744.05ms → 768.47ms (+3.3%)      | 14.00% → 14.96% (+6.9%)      |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.prepare_pages             | 13 → 13 (+0.0%)            | 47.52ms → 53.80ms (+13.2%)      | 67.01ms → 81.72ms (+22.0%) ⚠️   | 617.72ms → 699.44ms (+13.2%)     | 11.62% → 13.62% (+17.2%)     |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| code_index.chunk.index_file              | 276 → 276 (+0.0%)          | 2.96ms → 2.22ms (-25.0%) 🚀     | 6.02ms → 3.20ms (-46.8%) 🚀     | 818.29ms → 612.79ms (-25.1%) 🚀  | 15.39% → 11.93% (-22.5%) 🚀  |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| code_index.extract.parser_artifact       | 276 → 276 (+0.0%)          | 2.69ms → 2.05ms (-23.8%) 🚀     | 5.79ms → 3.13ms (-45.9%) 🚀     | 742.35ms → 566.08ms (-23.7%) 🚀  | 13.96% → 11.02% (-21.1%) 🚀  |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| query.artifact.batch.parallel_prepare    | 13 → 13 (+0.0%)            | 37.35ms → 43.34ms (+16.0%)      | 52.49ms → 68.55ms (+30.6%) ⚠️   | 485.61ms → 563.42ms (+16.0%)     | 9.13% → 10.97% (+20.2%) ⚠️   |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| 🆕 code_index.build.assemble             | 0 → 2 (+100.0%) ⚠️         | 0.00ns → 188.97ms (+100.0%) ⚠️  | 0.00ns → 203.55ms (+100.0%) ⚠️  | 0.00ns → 377.94ms (+100.0%) ⚠️   | 0.00% → 7.36% (+100.0%) ⚠️   |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| code_index.chunk.build                   | 276 → 276 (+0.0%)          | 1.92ms → 1.35ms (-29.7%) 🚀     | 4.72ms → 1.95ms (-58.7%) 🚀     | 530.02ms → 373.01ms (-29.6%) 🚀  | 9.97% → 7.26% (-27.2%) 🚀    |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+
| 🗑️ code_index.chunk.identify             | 276 → 0 (-100.0%) 🚀       | 1.89ms → 0.00ns (-100.0%) 🚀    | 4.71ms → 0.00ns (-100.0%) 🚀    | 521.63ms → 0.00ns (-100.0%) 🚀   | 9.81% → 0.00% (-100.0%) 🚀   |
+------------------------------------------+----------------------------+---------------------------------+---------------------------------+----------------------------------+------------------------------+

---

Generated with hotpath-rs

[ScriptedAlchemy]

Review of record (integration lane, base 7f1f8e02c).

The base moved under this PR: 1f7e1f0d5 applied blanket #[hotpath::measure_all] and marked trivial functions #[hotpath::skip]; 3e26c7596 then removed every measure_all ("retain only labeled function spans"). ~43 of the functions this PR instruments still carry that now-inert skip marker, which is the source of the 43 merge conflicts. Labeled semantic spans are exactly what the base retains, so the PR is policy-compatible; conflicts resolve to the PR side.

Two hunks are rejected and will be reverted in a follow-up commit on the merged head:

1. crates/tracedecay/src/host_admission.rs (14 sites) — 13 spans are on HostAdmissionTestRuntimeV1 (#[doc(hidden)], *_for_test methods) and one on the process background-CPU test fallback. That is test-fixture plumbing, not a production hot path; it inflates the label set with spans no operator journey crosses.
2. mcp.server.read_tokens_baseline — hotpath::future! around cg.get_tokens_saved(), whose body this same PR measures as daemon.store_meta.read_tokens_saved. Nested identical measurement of one await (same class as the doubled global_db.schema.persist.converge label in perf(global-db): bound schema convergence after project open #826).

Everything else (session_registry, engine, http_application, maintenance, pr_autotrack, scheduler, project_route, remaining server/ledger, profile_registry, store_meta) is accepted. hotpath::future! inside the Box::pin bodies in maintenance.rs is boxed so it adds no query depth; no target is a const fn.

The 9 red checks here are from a merge commit against a base that is ≥2 days stale (Format/Clippy/Test lanes fail on the base of that era); they will be re-judged on the merged #707 head. Gates before push: cargo check -p tracedecay --no-default-features --locked (production shape unaffected), --features hotpath check, clippy -D warnings, fmt, MCP suite ≥ 382/385.

[ScriptedAlchemy]

Integrated as merge commit 89e9e7711adfc2ff3473c6bfe006efaf997e4ed7 on codex/tracedecay-total-redesign-plan-reopened (current verified tip d4456c211c4be00503d7f32bdc6ba4c2ce81f953).

Evidence:

• unresolved review threads: 0
• all 102 intended PR feat(hotpath): fill daemon mcp/host-admission/io gaps #789 Hotpath labels retained after conflict resolution and latest-target integration
• cargo check -p tracedecay --features hotpath --locked: passed on final tip
• cargo check -p tracedecay --no-default-features --locked: passed on final tip
• git diff --check: passed
• full lib gate: merge 1513 passed / 39 failed versus clean target 1512 passed / 40 failed; the sole merge-only broad-suite failure passed exact on both target and merge, confirming baseline/non-deterministic suite failures rather than a PR feat(hotpath): fill daemon mcp/host-admission/io gaps #789 regression

The Hotpath future wrappers only wrap polling; newer target bodies, typed outcomes, lock scopes, cancellation paths, and daemon/MCP lifecycle ownership were retained.

[ScriptedAlchemy]

Follow-up on the review above, now landed on #707 as acf25b232 refactor(hotpath): drop test-fixture and duplicate spans from #789 (fast-forward on 77be8b23c):

• crates/tracedecay/src/host_admission.rs is byte-identical to its pre-feat(hotpath): fill daemon mcp/host-admission/io gaps #789 state (7f1f8e02c): the 14 daemon.host_admission.* spans on HostAdmissionTestRuntimeV1 / the background-CPU test fallback are gone and the base's #[hotpath::skip] markers are restored.
• mcp.server.read_tokens_baseline (hotpath::future! around cg.get_tokens_saved()) reverted to the plain await; daemon.store_meta.read_tokens_saved remains the single span for that read.
• The other 87 labels from this PR count exactly once each in the tree.

Gates on the integrated head: cargo fmt --check, cargo check -p tracedecay --no-default-features --locked (production shape), cargo check -p tracedecay -p tracedecay-store-runtime --features hotpath --locked (no query-depth overflow), cargo clippy -p tracedecay -p tracedecay-store-runtime -- -D warnings — all pass. MCP suite: 384/385 (single failure is context_relevance_eval_test, a retrieval-eval floor on the base, unrelated to instrumentation).