vietreport-admin/solution-erp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

wal: flush 20260814T1648

Browse Source
This commit is contained in:
pqhuy1987
2026-08-14 16:48:11 +07:00
parent a01147e9df
commit 85f240b2ed
8 changed files with 457 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
.claude/WAL.md
View File

@ -1,5 +1,5 @@
# WAL — auto-generated, không sửa tay
updated: 2026-08-14 | session: S196 (phiên-LOGIC L17, window 1) | branch: main
updated: 2026-08-14 | session: S197 (phiên-LOGIC L17, window 2 — nối `/tiep` 16:41) | branch: main
goal: bookend @open L17 chạy trọn 7 vai → owner YC-037 (chuẩn thuận + vá máy) & YC-038 (disposition) THI HÀNH XONG → dừng chờ chủ dự án.
@ -17,8 +17,9 @@ chain:
- [ ] 2 nhãn chờ chủ dự án phán: `carry-age-vacuous` (thước tuổi đảo dấu) · `ctx-t9-dogfood` (đang thu-hẹp)
- [ ] 17 slug ĐÓNG chưa gạch khỏi khối `HANDOFF:44`
- [ ] nghi-thức RE-STAMP chưa nhận đủ 5 dạng dấu đóng ⇒ kỳ sau lại phình
- [ ] trio-memory (eval→refine→audit) CHƯA chạy — luật định vị giữa 2 bookend, cửa-sổ này chưa tới
- [ ] `harness-refine-return.md` MISSING (nợ từ S195) · `grep.exe.stackdump` + 2 dir rỗng chưa dọn
- [!] trio-memory AUTO đang chạy @S197 (giữa 2 bookend L17, đúng vị-trí luật §2.1.9) — run `runs/2026-08-14-S197-trio-memory/`
- [x] dọn `grep.exe.stackdump` (untracked, xóa @S197); "2 dir rỗng" đo lại = 1 (`.vs/sd` IDE-local, giữ — VS tự tái tạo)
- [ ] `harness-refine-return.md` nợ S195 — mất theo phiên chết, không tái tạo; trio @S197 scribe đủ 3 return MỚI
next: CHỜ CHỦ DỰ ÁN — trả 2 nhãn ở trên, hoặc giao việc mới. KHÔNG tự mở việc.

5
.claude/agent-memory/.mfe-state.json
View File

@ -173,6 +173,11 @@
"strikes_total": 25,
"rca_entries": 16,
"at": "2026-08-14 10:32"
},
{
"strikes_total": 25,
"rca_entries": 16,
"at": "2026-08-14 16:46"
}
]
}

5
.claude/sessions/session-17/_tiep-1.md Normal file
View File

@ -0,0 +1,5 @@
ts: 2026-08-14T16:41:34+07:00
head-sha: a01147e9dfabda342abee48e34d920b1457ec064
window-ordinal: 2
jsonl-hint: D--Dropbox-CONG-VIEC-SOLUTION-SOLUTION-ERP/5a3a0802-1e24-486c-854f-ba08ebcd7ebe
account-label: none

81
.claude/workflows/runs/2026-08-14-S197-trio-memory/instrument-archive-gate.txt Normal file
View File

@ -0,0 +1,81 @@
============================================================
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
------------------------ --------- ----- ---------- ------- ------------ -------
back-end-reviewer-style 7057 no 0 0 - ok
cicd-monitor 19120 no 20 0 - ok
ctx-audit 30150 YES 18 1 ~20190 WATCH (strike 1<2): re-run; propose only after 2 consecutive over-cap
[D3 value-gate] 2 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
ctx-curator 17484 no 0 0 - ok
ctx-verifier 26910 YES 5 1 ~26910 WARN keep-floor hit (5); cannot auto-drain - SPLIT/condense entries by hand
[D3 value-gate] 1 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
database-agent 9038 no 6 0 - ok
database-reviewer-style 5824 no 1 0 - ok
frontend-designer 21090 no 8 0 - ok
front-end-reviewer-style 5913 no 1 0 - ok
harness-audit 25343 no 11 0 - ok
harness-eval 23967 no 12 0 - ok
harness-refine 24098 no 12 0 - ok
harvest-curator 24796 no 10 0 - ok
implementer-backend 22710 no 27 0 - ok
implementer-frontend 20846 no 13 0 - ok
investigator-api 9021 no 10 0 - ok
investigator-codebase 24567 no 12 0 - ok
lead-gap-auditor 32875 YES 12 1 ~21833 WATCH (strike 1<2): re-run; propose only after 2 consecutive over-cap
[D3 value-gate] 2 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
lead-stale-auditor 30710 YES 11 1 ~20419 WATCH (strike 1<2): re-run; propose only after 2 consecutive over-cap
[D3 value-gate] 1 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
office-document 4518 no 4 0 - ok
reviewer 28804 YES 12 1 ~24633 WATCH (strike 1<2): re-run; propose only after 2 consecutive over-cap
[D3 value-gate] 6 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
ring1-audit 27606 YES 11 1 ~14022 WATCH (strike 1<2): re-run; propose only after 2 consecutive over-cap
ring2-audit 34510 YES 13 2 ~23390 PROPOSE archive (strike 2>=2): move 7 lowest-value (position tiebreak) -> curate L1->L2 by hand
[D3 value-gate] 1 value-protected logical entr(y/ies) HARD-SKIPPED from the drain set (kept regardless of position/age); move-set = lowest-value only. NOTE: 9-token grep = lower-bound; em-main must value-scan the WHOLE proposed set (paraphrase/spine leak remains).
ring4-audit 3349 no 2 0 - ok
ring5-audit 6299 no 4 0 - ok
score-count-auditor 5637 no 2 0 - ok
score-counter 5682 no 3 0 - ok
test-specialist 20521 no 8 0 - ok
tooling-auditor 23022 no 12 0 - ok
[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 + 14 archive file(s)
-> PASS pointers 112 resolved 112 failed 0
[frontend-designer] _INDEX.md + 2 archive file(s)
-> PASS pointers 9 resolved 9 failed 0
[harvest-curator] _INDEX.md + 5 archive file(s)
-> PASS pointers 48 resolved 48 failed 0
[implementer-backend] _INDEX.md + 14 archive file(s)
-> PASS pointers 43 resolved 43 failed 0
[implementer-frontend] _INDEX.md + 4 archive file(s)
-> PASS pointers 44 resolved 44 failed 0
[investigator-codebase] _INDEX.md + 14 archive file(s)
-> PASS pointers 56 resolved 56 failed 0
[lead-gap-auditor] _INDEX.md + 3 archive file(s)
-> PASS pointers 8 resolved 8 failed 0
[lead-stale-auditor] _INDEX.md + 2 archive file(s)
-> PASS pointers 1 resolved 1 failed 0
[reviewer] _INDEX.md + 10 archive file(s)
-> PASS pointers 57 resolved 57 failed 0
[ring1-audit] _INDEX.md + 2 archive file(s)
-> PASS pointers 1 resolved 1 failed 0
[test-specialist] _INDEX.md + 4 archive file(s)
-> PASS pointers 33 resolved 33 failed 0
[tooling-auditor] _INDEX.md + 5 archive file(s)
-> PASS pointers 20 resolved 20 failed 0
------------------------------------------------------------
A7 GATE PASS - total pointers 432, resolved 432, failed 0
------------------------------------------------------------

45
.claude/workflows/runs/2026-08-14-S197-trio-memory/instrument-crystallized.txt Normal file
View File

@ -0,0 +1,45 @@
============================================================
crystallized-backfill.ps1 - measure-then-fill (DRY planner)
tier : lead
budget.json : D:\Dropbox\CONG_VIEC\SOLUTION\SOLUTION_ERP\.claude\agent-memory\memory-budget.json
cb config : present (target + hotload_sources read LIVE)
============================================================
### hot-load sources (LIVE byte-measure, UTF8)
source bytes state
-------------------------------------------- ---------- --------
docs/STATUS.md 256351 ok
docs/HANDOFF.md 246513 ok
.claude/governance/ACTIVE-MARKS.md 36153 ok
docs/changelog/migration-todos.md 90161 ok
docs/gotchas.md 137768 ok
.claude/agents/README.md 63637 ok
TOTAL 830583
### budget numbers (Upgrade-1) - ALL live-read, 0 hardcoded
Tran tong (live cap) : 380000 tok
hotload measured : 830583 bytes
tok RANGE [low..high] : [207646 .. 276861] tok (bytes/4 .. bytes/3.0)
measured_headroom : 103139 tok (cap - tok_high, worst-case)
target (owner) : 210000 tok
expected_backfill : 103139 tok (min(target,headroom)=headroom => headroom binds (target exceeds room))
### value-gated backfill source-order (planner / DRY - no pour)
precondition MET (target>0 AND headroom>0): would pull up to 103139 tok in order:
1. gist  archive/*.gist.md (4-field distilled)
2. value-marked  archive lines carrying gotcha# / anti-pattern / root-cause
3. curated RAG  search_memory high-value hits
(each stage DEDUP vs the hot-load sources above; stop at expected_backfill)
------------------------------------------------------------
HONEST CAVEAT
char/4 is NOT a real token count; VN-diacritic hot-memory ~3.0-3.5 byte/tok.
Headroom is a RANGE [bytes/4 .. bytes/3.0]; this planner uses bytes/3.0 (worst).
backfill defaults to 0 = TAT until the OWNER sets target>0 AND measured headroom>0.
Every budget number is the OWNER's authority; this script is READ-ONLY (never tunes/writes).
HEADROOM is measured over a STABLE FILE-FLOOR (the configured hotload_sources) = an UPPER BOUND.
The PEAK hot-feed also loads session-variable task-context NOT in this list; real headroom is SMALLER.
------------------------------------------------------------

239
.claude/workflows/runs/2026-08-14-S197-trio-memory/instrument-measure.txt Normal file
View File

@ -0,0 +1,239 @@
{
"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": "back-end-reviewer-style",
"l1_hot_bytes": 7057,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "cicd-monitor",
"l1_hot_bytes": 19120,
"l2_verbatim_bytes": 277739,
"l2_index_bytes": 24568,
"l2_gist_bytes": 122471,
"l1_over_30kb": false
},
{
"agent": "ctx-audit",
"l1_hot_bytes": 30150,
"l2_verbatim_bytes": 11521,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "ctx-curator",
"l1_hot_bytes": 17484,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "ctx-verifier",
"l1_hot_bytes": 26910,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "database-agent",
"l1_hot_bytes": 9038,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "database-reviewer-style",
"l1_hot_bytes": 5824,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "frontend-designer",
"l1_hot_bytes": 21090,
"l2_verbatim_bytes": 19050,
"l2_index_bytes": 2435,
"l2_gist_bytes": 3231,
"l1_over_30kb": false
},
{
"agent": "front-end-reviewer-style",
"l1_hot_bytes": 5913,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "harness-audit",
"l1_hot_bytes": 25343,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "harness-eval",
"l1_hot_bytes": 23967,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "harness-refine",
"l1_hot_bytes": 24098,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "harvest-curator",
"l1_hot_bytes": 24796,
"l2_verbatim_bytes": 92239,
"l2_index_bytes": 8325,
"l2_gist_bytes": 20317,
"l1_over_30kb": false
},
{
"agent": "implementer-backend",
"l1_hot_bytes": 22710,
"l2_verbatim_bytes": 94752,
"l2_index_bytes": 10487,
"l2_gist_bytes": 84345,
"l1_over_30kb": false
},
{
"agent": "implementer-frontend",
"l1_hot_bytes": 20846,
"l2_verbatim_bytes": 47796,
"l2_index_bytes": 5844,
"l2_gist_bytes": 44764,
"l1_over_30kb": false
},
{
"agent": "investigator-api",
"l1_hot_bytes": 9021,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "investigator-codebase",
"l1_hot_bytes": 24567,
"l2_verbatim_bytes": 171476,
"l2_index_bytes": 12448,
"l2_gist_bytes": 107741,
"l1_over_30kb": false
},
{
"agent": "lead-gap-auditor",
"l1_hot_bytes": 32875,
"l2_verbatim_bytes": 23458,
"l2_index_bytes": 1362,
"l2_gist_bytes": 7112,
"l1_over_30kb": true
},
{
"agent": "lead-stale-auditor",
"l1_hot_bytes": 30710,
"l2_verbatim_bytes": 9273,
"l2_index_bytes": 223,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "office-document",
"l1_hot_bytes": 4518,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "reviewer",
"l1_hot_bytes": 28804,
"l2_verbatim_bytes": 253991,
"l2_index_bytes": 13941,
"l2_gist_bytes": 69925,
"l1_over_30kb": false
},
{
"agent": "ring1-audit",
"l1_hot_bytes": 27606,
"l2_verbatim_bytes": 6761,
"l2_index_bytes": 682,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "ring2-audit",
"l1_hot_bytes": 34510,
"l2_verbatim_bytes": 8771,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": true
},
{
"agent": "ring4-audit",
"l1_hot_bytes": 3349,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "ring5-audit",
"l1_hot_bytes": 6299,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "score-count-auditor",
"l1_hot_bytes": 5637,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "score-counter",
"l1_hot_bytes": 5682,
"l2_verbatim_bytes": 0,
"l2_index_bytes": 0,
"l2_gist_bytes": 0,
"l1_over_30kb": false
},
{
"agent": "test-specialist",
"l1_hot_bytes": 20521,
"l2_verbatim_bytes": 37615,
"l2_index_bytes": 6170,
"l2_gist_bytes": 33811,
"l1_over_30kb": false
},
{
"agent": "tooling-auditor",
"l1_hot_bytes": 23022,
"l2_verbatim_bytes": 93802,
"l2_index_bytes": 3672,
"l2_gist_bytes": 18760,
"l1_over_30kb": false
}
]
}

60
.claude/workflows/runs/2026-08-14-S197-trio-memory/instrument-mfe.txt Normal file
View File

@ -0,0 +1,60 @@
=== MFE (Memory-Fidelity-EVAL) - Harness-16 - 2026-08-14 ===
(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)=11 AS-classes=15 net-extra-guards=5 net-extra-recurring-gotchas=7
=> DENOMINATOR = 38 items (raw before dedup: marks 11 + AS 15 + guards 16 + recurring 8)
[LEAD] Coverage-FIT (does the set FIT the hot-feed cap?):
span bytes=30945 => ~7736 - 10315 tok (RANGE, see caveat) / cap 380000 tok = 2.71% 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=11; >30d old but still Active=9 -> 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=25 RCA_entries=16 AS-classes=15
(rule: if strikes rise across runs while coverage stays high -> the score lies. warn-baseline=25 last_run_strikes=25)
[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)
back-end-reviewer-style N/A (no numbered anti-pattern / NEVER block in role-file to extract)
cicd-monitor 10/10 = 100% (MEASURED, not target-then-force)
ctx-audit 10/10 = 100% (MEASURED, not target-then-force)
ctx-curator 6/6 = 100% (MEASURED, not target-then-force)
ctx-verifier 7/7 = 100% (MEASURED, not target-then-force)
database-agent 7/7 = 100% (MEASURED, not target-then-force)
database-reviewer-style N/A (no numbered anti-pattern / NEVER block in role-file to extract)
front-end-reviewer-style 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 7/7 = 100% (MEASURED, not target-then-force)
implementer-backend 17/18 = 94% (MEASURED, not target-then-force)
implementer-frontend 12/13 = 92% (MEASURED, not target-then-force)
investigator-api 7/7 = 100% (MEASURED, not target-then-force)
investigator-codebase 8/8 = 100% (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 6/6 = 100% (MEASURED, not target-then-force)
reviewer 8/8 = 100% (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)
ring5-audit N/A (no numbered anti-pattern / NEVER block in role-file to extract)
score-count-auditor N/A (no numbered anti-pattern / NEVER block in role-file to extract)
score-counter N/A (no numbered anti-pattern / NEVER block in role-file to extract)
test-specialist 6/6 = 100% (MEASURED, not target-then-force)
tooling-auditor 8/8 = 100% (MEASURED, not target-then-force)
[SUB-WORKFLOW] N/A by design: fan-out workflow agents (hmw.js sub-<role>-<i>) 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) ===

18
.claude/workflows/runs/2026-08-14-S197-trio-memory/run.md Normal file
View File

@ -0,0 +1,18 @@
# run — trio-memory AUTO @S197 (phiên-LOGIC L17, window 2)
> Vòng AUTO 1-vòng/phiên GIỮA 2 bookend (owner @S151 "cứ đúng luật là tự làm"; nhà canonical `session-start.md §2.1.9`). L17 chưa chạy vòng nào → cửa sổ này đúng vị-trí. Dòng trạng-thái đã in @16:4x: `trio-memory: auto-vong @S197 (giua 2 bookend)`.
## taskList snapshot
1. script-arm: 4 instrument in số → `instrument-{measure,mfe,archive-gate,crystallized}.txt` — [x] 4/4 exit-0, 23.390 B
2. `harness-eval` (con ĐO) — PIN 4 instrument + memory-budget.json + NHIP-NO @S197 — [!] spawned
3. `harness-refine` (con ĐỀ-XUẤT) — PIN `harness-eval-return.md` + reinject-ledger — [ ] chờ nấc 1
4. `harness-audit` (con KIỂM) — PIN run-folder này — [ ] chờ nấc 2
5. Lead scribe return TỪNG con ngay khi trả (`harness-{eval,refine,audit}-return.md`) — 3 vai không Write/Edit
6. Post-wave assertion CÙNG TURN: `git status --short` kỳ vọng 0 tracked-change ngoài run-folder (§2.1.9(f) — lỡ boundary thì soát `git log --name-only` trên `wal:` commit)
## Ràng buộc
- TUẦN TỰ eval→refine→audit, CẤM song song (§2.1.9(b)).
- Mirror C1: bề-mặt-đo (agent-memory corpus) ĐÓNG BĂNG tới khi audit trả verdict; việc ghi phát sinh xếp hàng SAU.
- Nợ S195: `harness-refine-return.md` vòng trước MISSING (mất theo phiên chết) — vòng này scribe đủ 3 return MỚI, nợ cũ khai không-đòi-được.