Skip to content

0707 fold: novel HIGH + REVISIT densify - #66

Merged
24601 merged 1 commit into
mainfrom
cursor/0707-fold-novel-high-revisit-3202
Sep 21, 2026
Merged

24601 merged 1 commit into
mainfrom
cursor/0707-fold-novel-high-revisit-3202

Conversation

@24601

@24601 24601 commented Sep 21, 2026

Copy link
Copy Markdown
Owner

Hourly 0707 HIGH fold off latest main after merged #65 (0551 / §141). Fresh PR. Do not reopen or amend PR #23#65. Does not bump 0.5.0.

What this folds

Novel HIGH first cards (live REST this hour):

  • Sac-Y/Jev-cu 524★: text-only next-click. Quote theirs: 只传文字,不传截图. Codex CU executes. Local policy gates. Text only, no screenshots.
  • kyotofin/tax-doc-classifier 322★ Apache-2.0: Quote theirs: 100% of our tax document corpus at $0.001 per page. TaxCalcBench 0 strict errors. Blank IRS 38 strict errors 5.05%. 34× cheaper and 6× faster. 100% of corpus theirs not Harbor.
  • fhshaik/typesafe-mario 319★: Quote theirs: The model does not receive screenshots. Game success ≠ calibrated Noul.
  • droidrun/mobile-jev 307★ MIT: Quote theirs: 21 seconds for 9 actions. A completed booking is not demonstrated. ≠ Friedjof/jev-mobile.
  • realZachi/pg-jev, kitze/skillbox, itsmostafa/typesafe-mcp, kitze/unclutter, standardagents/jevpilot, plus named remainder (yoshi, pi-jev-auto-mode, snake, jev-code, calibrate, sarvam, winnow, advocaat, pg_typesafe, jev-chat, jev-browser, …).

REVISIT densify (keep prior section ids; no sibling first sightings):

  • AbdelStark/awesome-typesafe-jev densify §141: HEAD d6ea2a0d6cf4 README 234ae59a0b16 (was a6a68b57888a / e47993484e3a). Quote theirs: The field guide to typed decisions. Independent community project. SHA move is not a replica.
  • hf:Praveenrajus/jev-bench HTTP 401 (was 200). Densify §107/§125 remainder. Do not invent card content. theirs not Harbor.
  • hf:wayfind/metask-jev-4b-policy-mix densify §134 sha 5ecdd272ab4a. ≠ metask-ai/metask-jev.
  • patryckalves/jev-no-enem first card: ENEM 2025 theirs not Harbor. 56.6% (103/182) theirs. ECE 0.078 theirs.

Skip-thin empty SHA: developerekene/System-One, holotwist/laya, wuzhiping/jev-laya. HF 401 also on s1lv3rj1nx openjev-{healthcare-router,heldout,mixture}. First-party 0551 SHA bumps are not densify REVISIT. Skip Archer.

IDs

notes.md §142 / composition 681–696 / findings #124 / archive research/archive/hourly/2026-09-21T13/.

Checks

  • uniqueness_gate.py ok (0707 lock in 21 overlays; CHANGELOG has no dump wall)
  • evaluate_decisions.py --self-test ok
  • revisit_fingerprints.py --self-test ok
  • invented_signal: false

Catalog ≠ endorsement. Third-party benches stay theirs.

Open in Web Open in Cursor 

Catalog Jev-cu text-only CU, tax-doc, mario, mobile-jev, and peers.
Densify AbdelStark §141, jev-bench 401, metask §134; first-card ENEM.
Third-party benches stay *theirs*. Do not reopen PR #23-#65.

Co-authored-by: Basit Mustafa <24601@users.noreply.github.com>
@cursor

cursor Bot commented Sep 21, 2026

Copy link
Copy Markdown

ADV_PASS. Adversarial review + testing of #66 (HEAD 94a7b2e59c7d8833edc804dc56e43b41dbb7b78d).

Honesty checked vs live READMEs at claimed SHAs: Sac-Y/Jev-cu 524★ HEAD e2cc92d731fa README SHA 3deeafbc870f quote theirs 只传文字,不传截图; Codex CU executes; local policy gates; text-only CU is not screenshots. kyotofin/tax-doc-classifier 322★ Apache-2.0 HEAD 3e95a77f763c README SHA 72c4f74b542e quote theirs 100% of our tax document corpus at $0.001 per page; TaxCalcBench 0 strict errors; blank IRS 38 strict errors 5.05%; 34× cheaper and 6× faster; 261 IRS forms; 100% of corpus theirs not Harbor (strict score is not a hard gate). fhshaik/typesafe-mario 319★ HEAD ca22449ed187 README SHA c489f9350414 quote theirs The model does not receive screenshots; game success ≠ calibrated Noul. droidrun/mobile-jev 307★ MIT HEAD 395fc222beac README SHA d257fed2c5f7 quote theirs 21 seconds for 9 actions; A completed booking is not demonstrated; ≠ Friedjof/jev-mobile. patryckalves/jev-no-enem first card HEAD 7f85f787e3d1 README SHA 237b308df062; ENEM 2025 Caderno Azul 56.6% (103/182) theirs; ECE 0.078 theirs not Harbor.

REVISIT densify kept prior section ids (no sibling first sightings; no reopened uniqueness walls): AbdelStark/awesome-typesafe-jev densify §141 HEAD d6ea2a0d6cf4 README SHA 234ae59a0b16 was a6a68b57888a / e47993484e3a; The field guide to typed decisions; Independent community project; SHA move is not a replica; fingerprints notes_section 141. hf:Praveenrajus/jev-bench HTTP 401 was 200; densify §107/§125 remainder; do not invent card content. hf:wayfind/metask-jev-4b-policy-mix densify §134 sha 5ecdd272ab4a README SHA c534ee82b141; ≠ metask-ai/metask-jev; notes_section 134. First-party 0551 SHA bumps are not densify REVISIT (kev fingerprint still 4f8110a3f862 last_look 11:51Z / notes_section 45; APUS archive sha moved this hour and was not folded). UNIQ_0551 / UNIQ_0445 strings identical to main. Soft judgment ≠ hard gate throughout. Archer still promised_not_landed (Hub 401). invented_signal false. No version bump. Do not reopen #23#65.

Tests (python3, all exit 0):

  • .agents/skills/augustus/scripts/uniqueness_gate.py → uniqueness-gate ok (0707 chars=2934, 21 overlays, 5 revisit overlays)
  • .agents/skills/augustus/scripts/evaluate_decisions.py --self-test → self-test ok
  • research/revisit_fingerprints.py --self-test → revisit-fingerprints self-test ok

IDs unique vs main (34edba4 / merged #65): §142 / items 681–696 / batch #124 (main owns §141 / 665–680 / #123). Occupancy steal FAIL-as-expected for §142, 681, 696, #124; this branch only. SKILL 0.5.0 frozen. CHANGELOG Unreleased has no uniqueness dump wall. CI build success. READY_TO_MERGE. Do not merge from this comment.

@24601
24601 marked this pull request as ready for review September 21, 2026 13:43
@24601
24601 merged commit 41fa40c into main Sep 21, 2026
1 check passed
@24601
24601 deleted the cursor/0707-fold-novel-high-revisit-3202 branch September 21, 2026 13:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants