1b85713bb6
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
14 KiB
instruments @S151 (lead-run truoc trio - script-arm in SO, eval DOC + so-baseline)
1. measure-agent-memory.ps1
{ "measured_at_note": "run date stamped by caller (Date.now unavailable in workflow scripts)", "soft_cap_l1_bytes": 30720, "autoinject_cap_bytes": 25600, "agents": [ { "agent": "cicd-monitor", "l1_hot_bytes": 19281, "l2_verbatim_bytes": 228990, "l2_index_bytes": 22634, "l2_gist_bytes": 40303, "l1_over_30kb": false }, { "agent": "database-agent", "l1_hot_bytes": 7388, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "frontend-designer", "l1_hot_bytes": 21492, "l2_verbatim_bytes": 6624, "l2_index_bytes": 1531, "l2_gist_bytes": 3231, "l1_over_30kb": false }, { "agent": "harness-audit", "l1_hot_bytes": 1513, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "harness-eval", "l1_hot_bytes": 1547, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "harness-refine", "l1_hot_bytes": 1611, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "harvest-curator", "l1_hot_bytes": 20933, "l2_verbatim_bytes": 67374, "l2_index_bytes": 0, "l2_gist_bytes": 20317, "l1_over_30kb": false }, { "agent": "implementer-backend", "l1_hot_bytes": 19449, "l2_verbatim_bytes": 75787, "l2_index_bytes": 10105, "l2_gist_bytes": 23079, "l1_over_30kb": false }, { "agent": "implementer-frontend", "l1_hot_bytes": 15951, "l2_verbatim_bytes": 25581, "l2_index_bytes": 0, "l2_gist_bytes": 11125, "l1_over_30kb": false }, { "agent": "investigator-api", "l1_hot_bytes": 8510, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "investigator-codebase", "l1_hot_bytes": 10596, "l2_verbatim_bytes": 147348, "l2_index_bytes": 12384, "l2_gist_bytes": 43854, "l1_over_30kb": false }, { "agent": "lead-gap-auditor", "l1_hot_bytes": 24185, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "lead-stale-auditor", "l1_hot_bytes": 10128, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "office-document", "l1_hot_bytes": 3323, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "reviewer", "l1_hot_bytes": 18047, "l2_verbatim_bytes": 155039, "l2_index_bytes": 13780, "l2_gist_bytes": 35098, "l1_over_30kb": false }, { "agent": "ring1-audit", "l1_hot_bytes": 4481, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "ring2-audit", "l1_hot_bytes": 9840, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "ring4-audit", "l1_hot_bytes": 1047, "l2_verbatim_bytes": 0, "l2_index_bytes": 0, "l2_gist_bytes": 0, "l1_over_30kb": false }, { "agent": "test-specialist", "l1_hot_bytes": 22140, "l2_verbatim_bytes": 18065, "l2_index_bytes": 0, "l2_gist_bytes": 5374, "l1_over_30kb": false }, { "agent": "tooling-auditor", "l1_hot_bytes": 20082, "l2_verbatim_bytes": 66747, "l2_index_bytes": 0, "l2_gist_bytes": 18760, "l1_over_30kb": false } ] }
2. mfe-eval.ps1
=== MFE (Memory-Fidelity-EVAL) - Harness-16 - 2026-07-25 === (coverage/retention eval; NOT memoryDelta-routing-fidelity H6.7) lead hot-feed cap (live) = 380000 tok | mem-sub cap = 60000 tok
[LEAD] must-remember denominator (deduped, merge_guard_to_as=True): marks(Active+High)=9 AS-classes=15 net-extra-guards=5 net-extra-recurring-gotchas=7 => DENOMINATOR = 36 items (raw before dedup: marks 9 + AS 15 + guards 15 + recurring 8)
[LEAD] Coverage-FIT (does the set FIT the hot-feed cap?): span bytes=28191 => ~7048 - 9397 tok (RANGE, see caveat) / cap 380000 tok = 2.47% worst-case FIT = PASS (worst-case fits; headroom huge - must-remember set is tiny vs cap) CAVEAT: char/4 is NOT real tokens; VN-diacritic ~3.0-3.5 byte/tok => byte/4 = upper-bound headroom. Reported as a RANGE [bytes/4 .. bytes/3.0]; FIT uses bytes/3.0 (worst). Real count inside the band.
[LEAD] age-band (FLAG only - age=false-proxy, mark RC-...10-29-11): marks with parseable date=9; >30d old but still Active=4 -> KEPT (status-driven, age-blind) (guards/gotchas carry session-refs not dates -> age = status-driven only; drop on status-change, never by age)
[LEAD] Goodhart anchor (NO self-grading - real recurring-error signal): error-ledger: strikes_total=23 RCA_entries=14 AS-classes=15 (rule: if strikes rise across runs while coverage stays high -> the score lies. last_run_strikes=23)
[SUB] per-role memory coverage (denominator = role-file anti-patterns/baseline; numerator = carried in diary): match = exact-id OR >=2 distinct content-words after stop-list (leading-verb-only = false-max, defeated)
cicd-monitor 10/10 = 100% (MEASURED, not target-then-force) database-agent N/A (no numbered anti-pattern / NEVER block in role-file to extract) frontend-designer 2/2 = 100% (MEASURED, not target-then-force) harness-audit 5/5 = 100% (MEASURED, not target-then-force) harness-eval 5/5 = 100% (MEASURED, not target-then-force) harness-refine 5/5 = 100% (MEASURED, not target-then-force) harvest-curator N/A (no numbered anti-pattern / NEVER block in role-file to extract) implementer-backend 17/18 = 94% (MEASURED, not target-then-force) implementer-frontend 11/13 = 85% (MEASURED, not target-then-force) investigator-api 6/7 = 86% (MEASURED, not target-then-force) investigator-codebase 7/8 = 88% (MEASURED, not target-then-force) lead-gap-auditor 6/6 = 100% (MEASURED, not target-then-force) lead-stale-auditor 6/6 = 100% (MEASURED, not target-then-force) office-document N/A (no numbered anti-pattern / NEVER block in role-file to extract) reviewer 7/8 = 88% (MEASURED, not target-then-force) ring1-audit 5/5 = 100% (MEASURED, not target-then-force) ring2-audit 5/5 = 100% (MEASURED, not target-then-force) ring4-audit 5/5 = 100% (MEASURED, not target-then-force) test-specialist 6/6 = 100% (MEASURED, not target-then-force) tooling-auditor N/A (no numbered anti-pattern / NEVER block in role-file to extract)
[SUB-WORKFLOW] N/A by design: fan-out workflow agents (hmw.js sub--) are EPHEMERAL (no persistent diary; their only memory channel is the lead-injected memory-pack -> source-vs-cap is meaningless).
=== MFE done (deterministic block; READ-ONLY on budget; exit 0) ===
3. memory-archive-gate.ps1 (DRY)
============================================================ memory-archive-gate.ps1 - Harness-11 PART-A mode : DRY-RUN (no writes at all) cap : 25600 bytes (autoinject_cap) low-water : 21760 bytes (A4 hysteresis drain target = ratio 0.85) keep-floor : 5 newest entries (A5) strike-need : 2 consecutive over-cap runs to PROPOSE (A6)
PASS 1 - hot-tier over-cap planner (FLAG ONLY, no moves)
sub bytes over? entries strike after-est resolve
cicd-monitor 19281 no 20 0 - ok database-agent 7388 no 6 0 - ok frontend-designer 21492 no 7 0 - ok harness-audit 1513 no 4 0 - ok harness-eval 1547 no 4 0 - ok harness-refine 1611 no 4 0 - ok harvest-curator 20933 no 7 0 - ok implementer-backend 19449 no 23 0 - ok implementer-frontend 15951 no 19 0 - ok investigator-api 8510 no 9 0 - ok investigator-codebase 10596 no 11 0 - ok lead-gap-auditor 24185 no 9 0 - ok lead-stale-auditor 10128 no 2 0 - ok office-document 3323 no 3 0 - ok reviewer 18047 no 16 0 - ok ring1-audit 4481 no 3 0 - ok ring2-audit 9840 no 3 0 - ok ring4-audit 1047 no 1 0 - ok test-specialist 22140 no 17 0 - ok tooling-auditor 20082 no 8 0 - ok
(no sub over cap - hot tier within auto-inject budget)
[A6] DRY-RUN: strike counters NOT persisted (run with -Apply to advance strikes)
PASS 2 - A7 archive-integrity gate (NO-API: grep + measure only)
[cicd-monitor] _INDEX.md + 9 archive file(s) -> PASS pointers 101 resolved 101 failed 0 [frontend-designer] _INDEX.md + 2 archive file(s) -> PASS pointers 0 resolved 0 failed 0 [implementer-backend] _INDEX.md + 7 archive file(s) -> PASS pointers 41 resolved 41 failed 0 [investigator-codebase] _INDEX.md + 9 archive file(s) -> PASS pointers 55 resolved 55 failed 0 [reviewer] _INDEX.md + 7 archive file(s) -> PASS pointers 56 resolved 56 failed 0
A7 GATE PASS - total pointers 253, resolved 253, failed 0
4. tham-chieu output phien nay (da chay o bootstrap)
NHIP-NO: kiem light 1/6 deep 8/15 ok | tran-bo-nho 0 over-cap | run-chua-gom 0 | pull-cach 3 ngay (LUU Y: light_at nay = 25 sau signal-write) SHARD-PROBE: pending=15 | nguong=15 => IM crystallized-backfill: hotload 576921B -> tok-high 192307 | headroom 187693 < target 210000 (carry #26)