Shared memory and context tools for agentic work.
Code Rooms
{
"schema": "m1nd-bug-hunt-report-v0",
"round_id": "bughunt-p-limit-mc0-sweep-20260517T211556Z",
"generated_at": "2026-05-17T21:32:39.306612+00:00",
"repo": "p-limit",
"source_commit": "9f52583119f0cb0d85c6fec600c94a21fd89d060",
"seeded_bug_count": 5,
"seeded_bug_ids": [
"options-object-default-rejects-on-clear",
"reject-on-clear-falsy-non-boolean-accepted",
"map-non-array-iterable-index-lost",
"limit-function-drops-arguments",
"infinite-concurrency-rejected"
],
"lanes_completed": 6,
"lanes_expected": 6,
"lanes": [
"lane_id": "audit-01",
"instruction_mode": "m1nd-mission-control",
"completed": true,
"result_schema": "m1nd-bug-hunt-audit-result-v0",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-01.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-01.jsonl",
"findings_count": 5,
"matched_seeded_bug_ids": [
"infinite-concurrency-rejected",
"reject-on-clear-falsy-non-boolean-accepted"
"matched_findings": [
"finding_index": 0,
"finding_title": "Options object defaults rejectOnClear to true despite public default false",
"seeded_bug_id": "options-object-default-rejects-on-clear"
},
"finding_index": 1,
"finding_title": "Falsey non-boolean rejectOnClear values bypass validation",
"seeded_bug_id": "reject-on-clear-falsy-non-boolean-accepted"
"finding_index": 2,
"finding_title": "limit.map drops the index for non-array iterables",
"seeded_bug_id": "map-non-array-iterable-index-lost"
"finding_index": 3,
"finding_title": "limitFunction drops caller arguments",
"seeded_bug_id": "limit-function-drops-arguments"
"finding_index": 4,
"finding_title": "Infinity concurrency is rejected",
"seeded_bug_id": "infinite-concurrency-rejected"
}
"seeded_recall_count": 5,
"seeded_recall_rate": 1.0,
"missed_seeded_bug_ids": [],
"extra_unadjudicated_findings_count": 0,
"event_count": 4,
"agent_event_count": 4,
"first_event_at": "2026-05-17T21:24:57+00:00",
"first_agent_event_at": "2026-05-17T21:24:57+00:00",
"last_agent_event_at": "2026-05-17T21:29:03+00:00",
"agent_wall_clock_seconds": 246.0,
"assignment_to_first_agent_event_seconds": 0.0,
"first_finding_event_elapsed_seconds": 222.0,
"first_seeded_finding_event_elapsed_seconds": 222.0,
"timestamped_event_count": 4,
"timestamped_agent_event_count": 4,
"m1nd_usage_count": 7,
"mission_control": {
"applicable": true,
"unavailable": false,
"loop_complete": true,
"direct_proof_switch_count": 2,
"coverage_sweep_count": 1,
"do_not_guardrail_count": 2,
"verified_claim_signal_count": 5,
"rejected_claim_signal_count": 1,
"mission_start_count": 1,
"mission_next_count": 6,
"mission_verify_count": 6,
"mission_close_count": 1,
"required_step_count": 4,
"completed_step_count": 4,
"adherence_rate": 1.0
"agent_testimony": "I worked only in the audit-01 lane workspace and wrote only the assigned result JSON plus event JSONL. I did not read operator-only artifacts and did not patch source files. Runtime probes used an in-memory import of index.js with a minimal Queue stub because the lane lacks installed dependencies and I avoided creating node_modules or package-lock files."
"lane_id": "audit-02",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-02.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-02.jsonl",
"finding_title": "Options-object construction silently enables rejectOnClear by default",
"finding_title": "Falsy non-boolean rejectOnClear values bypass validation",
"finding_title": "limit.map drops mapper indexes for non-array iterables",
"finding_title": "limitFunction no longer forwards call arguments",
"finding_title": "Infinity is rejected as a concurrency limit",
"event_count": 6,
"agent_event_count": 5,
"first_event_at": "2026-05-17T21:15:56.209225+00:00",
"first_agent_event_at": "2026-05-17T21:16:00+00:00",
"last_agent_event_at": "2026-05-17T21:21:38+00:00",
"agent_wall_clock_seconds": 338.0,
"assignment_to_first_agent_event_seconds": 3.791,
"first_finding_event_elapsed_seconds": 210.0,
"first_seeded_finding_event_elapsed_seconds": 210.0,
"timestamped_event_count": 6,
"timestamped_agent_event_count": 5,
"m1nd_usage_count": 5,
"verified_claim_signal_count": 1,
"mission_next_count": 5,
"mission_verify_count": 2,
"agent_testimony": "I stayed in audit-02, did not read operator-only artifacts, did not patch source files, used the required /Users/kle1nz/m1nd/target/release/m1nd-mcp binary through probe_m1nd.py, and wrote only the assigned result JSON plus event JSONL. The lane workspace was briefly absent at the start, but appeared before direct proof; npm test could not run because dev dependencies were absent."
"lane_id": "audit-03",
"instruction_mode": "m1nd-trained",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-03.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-03.jsonl",
"finding_title": "limitFunction drops all caller arguments",
"finding_title": "limit.map omits mapper indexes for non-array iterables",
"finding_title": "Options object silently flips rejectOnClear default to true",
"finding_title": "Number.POSITIVE_INFINITY concurrency support was removed",
"event_count": 5,
"first_event_at": "2026-05-17T21:16:00+00:00",
"last_agent_event_at": "2026-05-17T21:28:18+00:00",
"agent_wall_clock_seconds": 738.0,
"first_finding_event_elapsed_seconds": 480.0,
"first_seeded_finding_event_elapsed_seconds": 480.0,
"timestamped_event_count": 5,
"m1nd_usage_count": 9,
"applicable": false,
"unavailable": null,
"loop_complete": null,
"direct_proof_switch_count": null,
"coverage_sweep_count": null,
"do_not_guardrail_count": null,
"verified_claim_signal_count": null,
"rejected_claim_signal_count": null,
"mission_start_count": null,
"mission_next_count": null,
"mission_verify_count": null,
"mission_close_count": null,
"required_step_count": null,
"completed_step_count": null,
"adherence_rate": null
"agent_testimony": "I did not read operator-only artifacts or patch source files. I worked only in the assigned audit-03 p-limit workspace for reads/probes and wrote only the assigned result JSON plus event JSONL outputs. The first helper path from the skill documentation was absent, so I used the benchmark-local m1nd operator helper at /Users/kle1nz/m1nd/skills/m1nd-operator/scripts/probe_m1nd.py with the requested --binary and --workspace-root flags."
"lane_id": "audit-04",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-04.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-04.jsonl",
"findings_count": 4,
"finding_title": "limit.map omits the index for non-array iterables",
"finding_title": "Options-object clearQueue rejects by default despite documented default false",
"finding_title": "Falsy non-boolean rejectOnClear values bypass runtime validation",
"seeded_recall_count": 4,
"seeded_recall_rate": 0.8,
"missed_seeded_bug_ids": [
"first_event_at": "2026-05-17T21:18:00+00:00",
"first_agent_event_at": "2026-05-17T21:18:00+00:00",
"last_agent_event_at": "2026-05-17T21:26:52+00:00",
"agent_wall_clock_seconds": 532.0,
"first_finding_event_elapsed_seconds": 370.0,
"first_seeded_finding_event_elapsed_seconds": 370.0,
"m1nd_usage_count": 8,
"agent_testimony": "Reran audit-04 in the assigned lane only. I did not read operator-only artifacts and did not patch source files. I wrote only the assigned lane result JSON and event JSONL. m1nd was used first, recovered from needs_ingest, then direct proof confirmed four concrete contract/behavior defects."
"lane_id": "audit-05",
"instruction_mode": "direct",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-05.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-05.jsonl",
"finding_title": "Options object changes rejectOnClear default from false to true",
"finding_title": "Number.POSITIVE_INFINITY concurrency is rejected",
"first_event_at": "2026-05-17T21:23:30+00:00",
"first_agent_event_at": "2026-05-17T21:23:30+00:00",
"last_agent_event_at": "2026-05-17T21:26:29+00:00",
"agent_wall_clock_seconds": 179.0,
"first_finding_event_elapsed_seconds": 130.0,
"first_seeded_finding_event_elapsed_seconds": 130.0,
"m1nd_usage_count": 0,
"agent_testimony": "Direct-mode audit only. I did not use m1nd tools, mission-control tools, m1nd helper scripts, or operator-only artifacts. I did not patch repository source files. Runtime probes used an in-memory minimal Queue replacement because dependencies were not installed in the lane workspace."
"lane_id": "audit-06",
"result_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/lane-results/audit-06.json",
"events_path": "/Users/kle1nz/m1nd/docs/benchmarks/bug-hunt-rounds/bughunt-p-limit-mc0-sweep-20260517T211556Z/event-streams/audit-06.jsonl",
"finding_title": "Options-object calls reject queued promises by default despite rejectOnClear defaulting to false",
"finding_title": "Falsy non-boolean rejectOnClear values are accepted at runtime",
"finding_title": "limitFunction drops every caller argument",
"extra_unadjudicated_findings_count": 1,
"event_count": 3,
"agent_event_count": 3,
"first_event_at": "2026-05-17T21:24:54.208000+00:00",
"first_agent_event_at": "2026-05-17T21:24:54.208000+00:00",
"last_agent_event_at": "2026-05-17T21:27:00.596000+00:00",
"agent_wall_clock_seconds": 126.388,
"first_finding_event_elapsed_seconds": 80.17,
"first_seeded_finding_event_elapsed_seconds": 80.17,
"timestamped_event_count": 3,
"timestamped_agent_event_count": 3,
"agent_testimony": "Worked only in the audit-06 lane workspace. Did not patch source files. Wrote only the assigned audit-06 result JSON and event JSONL outputs. Direct import/test execution was blocked by missing node_modules/yocto-queue, so behavioral checks used a no-write in-memory Queue stub while preserving current index.js logic."
"arms": {
"direct": {
"lane_count": 2,
"completed_lane_count": 2,
"seeded_bug_count_per_lane": 5,
"seeded_recall_total": 9,
"seeded_possible_total": 10,
"seeded_recall_rate": 0.9,
"per_lane_seeded_recall_counts": [
5,
4
"median_seeded_recall_count": 4.5,
"average_seeded_recall_count": 4.5,
"median_agent_wall_clock_seconds": 152.694,
"median_first_finding_event_elapsed_seconds": 105.085,
"median_first_seeded_finding_event_elapsed_seconds": 105.085,
"total_findings": 10,
"extra_unadjudicated_findings_total": 1,
"mission_control_loop_complete_lanes": 0,
"mission_control_unavailable_lanes": 0,
"median_mission_next_count": null,
"median_direct_proof_switch_count": null,
"median_coverage_sweep_count": null,
"median_mission_control_adherence_rate": null,
"audit-05",
"audit-06"
]
"m1nd-mission-control": {
"seeded_recall_total": 10,
5
"median_seeded_recall_count": 5.0,
"average_seeded_recall_count": 5.0,
"median_agent_wall_clock_seconds": 292.0,
"median_first_finding_event_elapsed_seconds": 216.0,
"median_first_seeded_finding_event_elapsed_seconds": 216.0,
"extra_unadjudicated_findings_total": 0,
"mission_control_loop_complete_lanes": 2,
"median_mission_next_count": 5.5,
"median_direct_proof_switch_count": 2.0,
"median_coverage_sweep_count": 1.0,
"median_mission_control_adherence_rate": 1.0,
"audit-01",
"audit-02"
"m1nd-trained": {
"median_agent_wall_clock_seconds": 635.0,
"median_first_finding_event_elapsed_seconds": 425.0,
"median_first_seeded_finding_event_elapsed_seconds": 425.0,
"total_findings": 9,
"audit-03",
"audit-04"
"comparability": {
"all_lane_results_present": true,
"primary_arm_lane_counts": {
"direct": 2,
"m1nd-mission-control": 2,
"m1nd-trained": 2
"mission_control_validity": {
"present": true,
"evaluable_lane_count": 2,
"partial_or_unavailable_lane_count": 0,
"missing_result_lane_count": 0,
"all_completed_lanes_evaluable": true,
"evaluable_lane_ids": [
"partial_or_unavailable_lane_ids": [],
"missing_result_lane_ids": [],
"non_claim": "Mission Control recall is not attributable to MC0 unless the lane is evaluable."
"rate_comparison_available": true,
"balanced_lane_counts": true,
"comparability_notes": [
"Compare rates rather than raw totals when arm lane counts differ.",
"Extra findings are unadjudicated and are not used as precision penalties."
"top_line": {
"median_seeded_recall_count": 4.5
"median_seeded_recall_count": 5.0
"invalidated_attempts": [
"disposition": "rerun_completed",
"reason": "lane started before seeded workspace was materialized"
"public_claim_worthy": false,
"public_claim_blockers": [
"single internal round",
"one fixture repo",
"seeded recall only; extra findings are not independently judged"
"non_claims": [
"Finding extra real issues is allowed but seeded recall is measured against these five defects.",
"Primary auditors are not told bug count or comparison arm.",
"This round tests p-limit seeded behavioral defects, not universal agent performance.",
"agent testimony is not evidence without scored finding artifacts",
"extra findings are reported as unadjudicated, not as false positives",
"m1nd does not replace tests, compiler output, git history, rg, or direct file truth",
"one bug-hunt round is not a public performance claim",
"seeded recall does not measure all real defects in the fixture repo"