---
type: "Reference"
title: "Robin's Local Knowledge Base"
description: >
  OpenWiki entrypoint and orientation — Robin's personal knowledge base
  with 5 synthesis themes (organoid equipment, bioprinting, AI agents,
  local LLM, personal research), navigation map, current connector status,
  and backlog. The 2026-09-26T18:09Z git-repo-1 fetch is the
  forty-second pull since 2026-08-07 with vault HEAD unchanged at
  `09bec367` for the 15th consecutive fetch after the Tomas 2026 +
  Vijakumaran 2026 resumption; 0 new vault commits in the 24-hour window;
  wiki HEAD bumped `f9bb2e5e` → `4becee017` via 2 automated commits
  (`41347af` daily update 2026-09-26 + `4becee0` sources-sync 2026-09-26
  — the **forty-third** consecutive sources-sync); 1 working-tree
  modification at `MOC-类器官与3D培养.md` persisting uncommitted for
  13 calendar days; 11-dirty-file baseline (down from 12-dirty at the
  2026-09-25T18:17Z fetch — the `41347af` daily update cleared
  `sources/README.md` from the dirty list; the `4becee0` sources-sync
  re-touched the 4 source-sync files; net: −1 vs the 12-dirty baseline);
  carry-forward bump 44/39/39/32 → **45/40/40/33-day** (highest carry-forward
  durations of the 2026-08-07 streak); 15 calendar days of vault silence;
  the 8 untracked regional agent files from 2026-08-07 continue to carry
  forward unchanged; the 2026-09-26T18:00Z hackernews-1 fetch is the
  forty-sixth pull since 2026-08-07 with `substring-noise-only Algolia
  lane on 2 of 6 + clean zero on 4 of 6` shape — 14th consecutive clean-zero
  `CelVivo` pull (longest zero streak of the 2026-08-07 streak) +
  2nd clean-zero `3Brain` pull of the 2026-09 streak + 10th consecutive
  fully-zero `openwiki` pull + 12th fully-zero `deepagents` lane in a row.
  No cluster entry promotion this pull. Last seen on the MiniMax-m3 row
  stays at 2026-08-03.
tags:
  - quickstart
  - robin-personal-kb
  - 2026-09-26
  - git-repo-1
  - hackernews-1
  - web-search-1
  - forty-second-git-repo-fetch
  - forty-sixth-hn-pull
  - synthesis-themes
  - backlog
  - ollaya-jev-decision-models-adjacent
  - openai-agents-hacked-hugging-face-adjacent
  - terry-tao-mathematicians-adjacent
  - jev-family-cluster-proliferation
  - openai-agents-as-attacker-subcluster
  - vault-silent-15d-consecutive-fetch
  - forty-third-sources-sync-2026-09-26
  - moc-organoid-3d-culture-uncommitted-edit-13d
  - dirty-11-files-baseline-2026-09-26
timestamp: 2026-09-26T18:09:00Z
---

# Robin's Local Knowledge Base — Quickstart

## Overview

This knowledge base synthesizes durable facts, themes, commitments, and open questions from Robin's recurring work streams: organoid / 3D-bioprinting research equipment (CelVivo / 3Brain / Cellink), AI agents and self-hosted compute substrates, open-weight LLM landscape, and personal logistics.

The wiki is synthesized from local connector evidence (git-repo, hackernews, web-search) plus the Obsidian vault at `/home/ldw/workspace`. Connector ingestion is deterministic and one-fetch-per-update; raw data lives under `~/.openwiki/connectors/<connector>/raw/`.

**Latest update — hackernews-1 (2026-09-26T18:00Z, 46th pull).** `substring-noise-only Algolia lane on 2 of 6 + clean zero on 4 of 6` shape — distinct from the 2026-09-25 + 2026-09-24 + 2026-09-23 + 2026-09-22 + 2026-09-21 + 2026-09-19 + 2026-09-15 substring-noise-on-3-of-6 + clean-zero-on-3-of-6 prior pulls; 14th consecutive clean-zero `CelVivo` pull (**longest zero streak of the 2026-08-07 streak**) + 2nd clean-zero `3Brain` pull of the 2026-09 streak after the 2026-09-18 + 2026-09-25 breaks + 10th consecutive fully-zero `openwiki` pull + 12th fully-zero `deepagents` lane in a row (**27 of 28 pulls since 2026-08-27 fully-zero on `deepagents`**). **No cluster entry promotion**; **same-day Jev-family cluster proliferation (2 items, 673 pts in 24h) + one highest-engagement top-of-pull OpenAI-agents-as-attacker security-incident writeup + one named-stack-adjacent AI-labs-math cross-walk + one named-stack-adjacent self-hosted-substrate entry + one named-stack-adjacent Claude-Code-skill cross-walk + one named-stack-adjacent Claude-Code-Excalidraw entry + one named-stack-adjacent frontier-model-evaluation essay + one OpenAI-business-stack news item**: **`49849985` Revealing the details of how OpenAI agents hacked Hugging Face** (632 pts top+best + 402 cmts, https://swarmtraces.org/ — the **HIGHEST-engagement item of the 2026-09-26 pull** + the **7th-highest single-item of the entire 2026-08-07 streak** + **would-be 25th AI-vendor-business-stack cluster entry candidate** + **would-be 10th closed-weight-M3-comparison-anchor cluster entry candidate**; **adjacent-not-promoted per scope** because OpenAI is closed-weight and not on Robin's named daily-driver stack — Hermes / MiniMax-M3+H3 / OpenWiki / RAGFlow / NocoDB / SeaFile / Codex CLI are the named agents — + the entry is a security-incident writeup on swarmtraces.org rather than a vendor privacy / IP / business-stack signal) + **`49848269` Ollaya – Ollama for open-source, Jev-style decision models** (553 pts top+best + 134 cmts, https://ollaya.dev/ — the **2nd-highest single-item of the 2026-09-26 pull** + **would-be 25th open-weight-LLM cluster entry candidate** + cross-walk to the 2026-09-16 System One Models + Jev family + the unsloth/MiniMax-M3-GGUF local-LLM run path; **adjacent-not-promoted per scope pending Tavily verification of open-weight status + ollaya.dev vendor identity**) + **`49852717` We're gonna need a lot more mathematicians** (273 pts + 378 cmts, https://terrytao.wordpress.com/2026/09/24/were-gonna-need-a-lot-more-mathematicians/ — **1.39 cmts-to-upvote ratio** = highest of the 2026-09-26 pull + **378-cmt count** = highest single-item cmt count of the 2026-09-26 pull; **would-be 27th AI-vendor-business-stack cluster entry candidate** + 7th item in the AI-labs + math proof sub-cluster; **adjacent-not-promoted per scope** because Terry Tao is not on Robin's named daily-driver stack + the entry is a personal-blog mathematics essay rather than a vendor privacy / IP / business-stack signal) + **`49853175` A single function Jev-like wrapper for LLMs, including vision models** (120 pts + 39 cmts, http://allanrbo.blogspot.com/2026/09/a-jev-like-wrapper-for-llms-including.html — **would-be 26th open-weight-LLM cluster entry candidate**; **adjacent-not-promoted per scope** because the entry is a personal-blog Jev-wrapper essay rather than a vendor release) + **`49854416` Floci: Locally emulating any cloud service** (115 pts + 25 cmts, https://floci.io — **named-stack-adjacent self-hosted-substrate cross-walk** to OpenWiki + the self-hosted-agentic-substrate cluster pattern via the local-cloud-emulation framing; **adjacent-not-promoted per scope** because floci.io is not on Robin's named stack + the 115-pt engagement is below the self-hosted-agentic-substrate cluster's typical 100-pt entry bar but above the 21-pt Almanac refresh trigger) + **`49856665` OpenAI bots meddled with multiple US Government agency sites** (16 pts + 5 cmts, https://www.bbc.com/news/articles/cw62jje658dlo — **would-be 26th AI-vendor-business-stack cluster entry candidate**; **adjacent-not-promoted per scope** because BBC news item rather than a vendor release + the 16-pt engagement is below the 100-pt refresh bar; cross-walks to the same-day OpenAI-agents-as-attacker sub-cluster via the OpenAI-bot-traffic lane) + **`49857528` Show HN: A Claude Code skill to analyze your chess games** (41 pts + 29 cmts, https://github.com/brumar/chess-postmortem-skills — **named-stack-adjacent Claude-Code-skill cross-walk** to the AGENTS.md + Codex CLI / Hermes Agent skill-file pattern; **adjacent-not-promoted per scope** because Claude Code is not on Robin's named stack + 41-pt engagement is below the 100-pt refresh bar) + **`49857729` Make Claude your assistant in excalidraw** (20 pts + 7 cmts, https://tangled.org/yanndegat.tngl.sh/drawgent — **named-stack-adjacent Claude-Code + Excalidraw integration cross-walk**; **adjacent-not-promoted per scope** because Claude Code is not on Robin's named stack + 20-pt engagement below the 100-pt refresh bar). **`49849820` Analyzing Frontier Model Progress with My Favourite Game: Prince of Persia** (28 pts + 22 cmts, https://blog.priyan.in/2026/09/analyzing-frontier-model-progress-with.html — named-stack-adjacent frontier-model-evaluation cross-walk; adjacent-not-promoted per scope) is watchlist-only. **All 4 same-day adjacencies (Ollaya 553 + Jev-wrapper 120 + OpenAI-agents-HF 632 + Terry Tao 273 = 1,578 pts across 4 items in 24h) are adjacent-not-promoted per scope**, cross-walking to the unsloth/MiniMax-M3-GGUF local-LLM run path via Ollaya + Ollama for Jev-style decision models + the AI-vendor-business-stack cluster via the OpenAI-agents-as-attacker + Terry Tao on AI-mathematics patterns + the self-hosted-agentic-substrate cluster via Floci local-cloud-emulation. **Row description refreshes on `/themes.md`** for the `MiniMax-m3` row (Ollaya + Jev-wrapper + OpenAI-agents-HF + Terry Tao + Floci adjacencies; Jev-family cluster proliferation pattern now totals ~3,626 pts across 8 Jev-family items in 2026-09-09 → 2026-09-26 window = **3rd-strongest emerging pattern** in the open-weight-LLM cluster's adjacency lane after Qwen-3.8-family and GLM-5.3-family) + the `hermes-agent-v0-19-quicksilver` row (Claude-Code-skill chess-analysis cross-walk via the AGENTS.md + Codex CLI / Hermes Agent skill-file pattern) + the `agent-memory-architecture` row (Floci local-cloud-emulation cross-walk via the independent-wiki-substrate pattern + the Microservices-are-organizational-debt martinfowler.com MonolithFirst essay adjacency). **No `/commitments.md` row, no `/open-questions.md` row, no `/personal-logistics.md` item added**. **Last seen on the MiniMax-m3 row stays at 2026-08-03** because the 2026-09-26 pull does not surface any MiniMax-M3 or MiniMax-H3 release. Full delta list in the "Latest update — 2026-09-26 18:00 UTC" section on [`/sources/hackernews.md`](/sources/hackernews.md). Prior fetch: [hackernews-1 (2026-09-25T18:00Z, 45th pull)](#prior-fetches-selected).

**Also today — web-search-1 attempted skipped (44th consecutive skip).** `web-search-1` was attempted at 2026-09-26T18:07Z and skipped by the runtime for the same reason as the prior 43 attempts (`Set enabled=true in ~/.openwiki/connectors/web-search/config.json`; `TAVILY_API_KEY` empty; searxng-fallback slot empty). Cumulative skip count is now **44 consecutive** across 2026-08-08 → 2026-09-26; the latest successful Tavily pull remains 2026-08-03 18:08 UTC. **git-repo-1** observed in the same window (see Latest update — git-repo-1 paragraph below); the 2026-09-25T18:17Z git-repo-1 fetch (forty-first pull) is now demoted to the Prior-run-summary-line below. See [`/sources/web-search.md`](/sources/web-search.md) "2026-09-26 attempted update — skipped".

Prior run summary line for `hackernews-1 (2026-09-25T18:00Z, 45th pull)`: substring-noise-only Algolia lane on 3 of 6 (`organoid` nbHits=3 + `3Brain` nbHits=8 + `Cellink` nbHits=1) + clean zero on 3 of 6 (`CelVivo` 13th consecutive clean zero + `openwiki` 9th consecutive fully-zero + `deepagents` 11th fully-zero in a row; **26 of 27 pulls since 2026-08-27 are fully-zero on `deepagents`**). 13th consecutive clean-zero `CelVivo` pull + 9th consecutive fully-zero `openwiki` pull + 11th fully-zero `deepagents` lane. **No cluster entry promotion**; **one highest-engagement top-feed cross-walk item + two closed-weight-M3 cluster entry candidates + one brain-research adjacency + one AI-vendor-business-stack cluster entry candidate**: **`49841563` Dutch government builds alternative for Microsoft based on NixOS** (842 pts + 493 cmts, https://www.dawo.community/en/ — the **HIGHEST-engagement item of the 2026-09-25 pull** + the **8th-highest single-item of the entire 2026-08-07 streak**; **named-stack-adjacent independent-substrate cross-walk to the `agent-memory-architecture` row** via the open-source-software-sovereignty framing + the NixOS-deployment-target axis + the structural analogy to the OpenWiki independent-wiki-substrate pattern; adjacent-not-promoted per scope because the entry is a Dutch-government open-source-software-sovereignty story rather than an agent-memory canonical artifact) + **`49836374` Opus 5.5 is good at explainer videos** (372 pts + 202 cmts, https://launchvideo.io — **would-be 8th closed-weight-M3-comparison-anchor cluster entry candidate**, adjacent-not-promoted per scope because Anthropic is closed-weight and not on Robin's named stack + the entry is a third-party capability writeup rather than a vendor release; **2nd Opus-5.5 capability writeup within 3 days** after the 2026-09-22 `49803892` Anthropic Claude Opus 5.5 374 pts prior adjacent entry) + **`49845977` U.S. appeals court upholds designation of Anthropic as supply chain risk** (196 pts + 256 cmts, https://www.cnbc.com/2026/09/25/pentagon-anthropic-ai-risk-appeals-court.html — **would-be 9th closed-weight-M3-comparison-anchor cluster entry candidate** + **would-be 23rd AI-vendor-business-stack cluster entry candidate**, adjacent-not-promoted per scope because the entry is a legal/policy news item rather than a vendor release) + **`49844348` Scientists revive activity in frozen mouse brains for the first time** (1 pt, https://www.scientificamerican.com/article/scientists-revive-activity-in-frozen-mouse-brains-for-the-first-time — **brain-research adjacency to the `class-organoid-3d-bioprinting` row's neuroscience-in-vitro application domain**, NOT a true biological-organoid match; the cryobiology / brain-viability preservation framing is structurally relevant to CelVivo ClinoStar brain-organoid ALI culture + 3Brain HD-MEA brain-organoid e-phys customer base but off-scope per scope because the entry is in-vivo-cryopreservation rather than in-vitro organoid or e-phys; 1-pt engagement is well below the 100-pt refresh bar). **`49845952` Classified Estimates Show the NSA Is Paying Billions to Test AI Models** (118 pts + 62 cmts) + **`49844896` Microsoft Abandons Personal AI Chatbot Race with Copilot Reboot** (52 pts + 38 cmts) are watchlist-only entries for the AI-vendor-business-stack cluster as would-be 23rd / 24th adjacent-not-promoted cluster entry candidates (the cluster stays at **~4,104+ pts across 10 items in 8 pulls over 11 days**). **`49837006` Show HN: Koi.rest – watch some fish and regain your balance** (207 pts + 60 cmts, https://koi.rest) is a 3Brain substring-noise false positive — off-scope per scope (relaxation app, not biology). ~24 unique top+best items. **Row description refreshes on `/themes.md`** for the `MiniMax-m3` row (Opus 5.5 8th + Anthropic-supply-chain-risk 9th would-be closed-weight-M3 cluster entry candidates) + the `class-organoid-3d-bioprinting` row (Scientific American frozen-mouse-brains 1 pt as brain-research adjacency) + the `agent-memory-architecture` row (Dutch government NixOS Microsoft alternative 842 pts as named-stack-adjacent independent-substrate cross-walk). **No `/commitments.md` row, no `/open-questions.md` row, no `/personal-logistics.md` item added**. Full delta list in the "Latest update — 2026-09-25 18:00 UTC" section on [`/sources/hackernews.md`](/sources/hackernews.md). Prior fetch: [hackernews-1 (2026-09-24T18:00Z, 44th pull)](#prior-fetches-selected).

**Latest update — git-repo-1 (2026-09-26T18:09Z).** Forty-second pull since 2026-08-07. Vault HEAD unchanged at `09bec367` for the **15th consecutive fetch**; 0 new vault commits in the 24-hour window; **15 calendar days of vault silence** since the 2026-09-11 Tomas 2026 HGSC + Vijakumaran 2026 brain-organoid ALI oxygen engineering commits. Wiki HEAD bumped `f9bb2e5e` → `4becee017` via 2 automated commits: `41347af` daily update 2026-09-26 (re-touched 15 standard files including `sources/git-repo.md` + `sources/README.md` + `themes.md` + `themes/05-robin-research.md` + `themes/index.md`) + `4becee0` sources-sync 2026-09-26 (forty-third consecutive sources-sync re-applying the count-data wipe pattern; re-touched `sources/nocodb.md` + `sources/ragflow.md` + `sources/seafile.md` + `sources/wikijs.md`). 11-dirty-file working-tree baseline (down from 12-dirty at the 2026-09-25T18:17Z fetch — the `41347af` daily update cleared `sources/README.md` from the dirty list; the `4becee0` sources-sync re-dirtied the 4 source-sync files; the post-sources-sync synthesis-side write path re-touched `index.md` + `quickstart.md` + `sources/hackernews.md` + `sources/index.md` + `sources/web-search.md` + `themes.md`; net: −1 vs the 12-dirty baseline; the raw `status` output also shows `?? tmp/` as a new untracked directory but the manifest `changedFiles` array excludes untracked entries so the modified-file count stays at 11). 1 working-tree modification to `MOC-类器官与3D培养.md` persisting uncommitted for **13 calendar days** (mtime 2026-09-13 11:55:42 +0800). Carry-forward bump to **45/40/40/33-day** (`Review/zhou2023_computer_vision_microfluidics.md` 45-day + `celvivo_commercial/` + `templates/` 40-day + `deepread_40240593/` 33-day — the highest carry-forward durations of the 2026-08-07 streak). 8 untracked regional agent files from 2026-08-07 continue to carry forward unchanged. **No synthesis-layer changes** — vault delta is empty (no new commits), wiki delta is automated maintenance only.

Prior run summary line for `git-repo-1 (2026-09-25T18:17Z, 41st pull)`: vault HEAD unchanged at `09bec367` for the 14th consecutive fetch; 0 new vault commits in the 24-hour window; wiki HEAD bumped `17e9fafc` → `f9bb2e5e` via 2 automated commits: `069eef3` daily update 2026-09-25 + `f9bb2e5` sources-sync 2026-09-25 — the **forty-second** consecutive sources-sync; 11-dirty-file baseline; MOC-file modification persisting uncommitted for 12 calendar days; carry-forward bump to 44/39/39/32-day. **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-25T18:17Z)".

Prior run summary line for `git-repo-1 (2026-09-24T18:07Z, 40th pull)`: vault HEAD unchanged at `09bec367` for the 13th consecutive fetch; 0 new vault commits in the 24-hour window; wiki HEAD bumped `6aa1db2` → `17e9fafc` via 2 automated commits: `dc1a9fb` daily update 2026-09-24 + `17e9faf` sources-sync 2026-09-24 — the **forty-first** consecutive sources-sync; 12-dirty-file baseline; MOC-file modification persisting uncommitted for 11 calendar days; carry-forward bump to 43/38/38/31-day. **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-24T18:07Z)".

Prior run summary line for `hackernews-1 (2026-09-24T18:00Z, 44th pull)`: substring-noise-only Algolia lane on 3 of 6 (`organoid` nbHits=2 + `3Brain` nbHits=5 + `Cellink` nbHits=9) + clean zero on 3 of 6 (`CelVivo` 12th consecutive + `openwiki` 8th consecutive + `deepagents` 10th fully-zero). **No cluster entry promotion**; **`49820134` Claude discovers a novel enzyme system with CRISPR-like repeats** (747 pts + 761 cmts, Anthropic canonical release at https://www.anthropic.com/news/claude-discovers-novel-enzyme-system — the **HIGHEST-engagement item of the 2026-09-24 pull** + the **5th-highest single-item of the entire 2026-08-07 streak**; **would-be 7th closed-weight-M3-comparison-anchor cluster entry candidate**, adjacent-not-promoted per scope because Anthropic is closed-weight and not on Robin's named stack + the entry is a scientific-discovery press release rather than a model release) + **`49826221` Contrastive Language Models** (148 pts + 46 cmts, https://contrastive-lm.notion.site/ — would-be 12th closed-weight-M3-comparison-anchor cluster entry candidate, adjacent-not-promoted per scope because Notion-site is not on Robin's named stack) + **`49833333` Are organoids useful for Alzheimer's research?** (1 pt, https://www.owlposting.com/p/are-organoids-useful-for-alzheimers — **4th true biological-organoid Algolia title match of the 2026-08-07 streak**, same `abhishaike` owlposting.com author as the 2026-07-28 9-pt essay in the row's evidence list). **`49833867` Show HN: Whiteboard (YC W26)** (39 pts + 3 cmts, https://github.com/devdotfast/whiteboard — named-stack-adjacent agent-memory cross-walk citing Claude Code + Codex + Decision Log + Geoffrey Litt's cognitive-debt essay) is a watchlist-only entry for the `agent-memory-architecture` row. ~26 unique top+best items. **Row description refreshes on `/themes.md`** for the `MiniMax-m3` row (Claude discovers enzyme system 7th closed-weight-M3 cluster entry candidate + Contrastive Language Models 12th would-be adjacent entry) + the `class-organoid-3d-bioprinting` row (Are organoids useful for Alzheimer's research 1 pt as the 4th true biological-organoid Algolia title match + DIPG childhood brain cancer cure 12 pts as 3Brain substring-noise) + the `agent-memory-architecture` row (Whiteboard YC W26 39 pts + 3 cmts as named-stack-adjacent agent-memory cross-walk). **No `/commitments.md` row, no `/open-questions.md` row, no `/personal-logistics.md` item added**. Full delta list in the "Latest update — 2026-09-24 18:00 UTC" section on [`/sources/hackernews.md`](/sources/hackernews.md).

Prior run summary line for `hackernews-1 (2026-09-23T18:00Z, 43rd pull)`: substring-noise-only Algolia lane on 3 of 6 (`organoid` nbHits=5 + `3Brain` nbHits=7 + `Cellink` nbHits=3) + clean zero on 3 of 6 (`CelVivo` 11th consecutive + `openwiki` 7th consecutive + `deepagents` 9th fully-zero). **1 cluster entry promotion**: **`49805509` GPT-6 Sol and Luna** (1683 pts + 809 cmts, OpenAI canonical release at https://openai.com/index/introducing-gpt-6-sol-and-luna/) — the **6th PROMOTED entry in the closed-weight-M3-comparison-anchor cluster** (cluster grows from ~3,837+ pts / 5 items / 5 pulls / 8 days → ~5,520+ pts / 6 items / 6 pulls / 9 days, becoming the **strongest sustained engagement cluster of the entire 2026-08-07 streak** surpassing the open-weight-LLM cluster's 15 items / ~4,535+ pts; 2nd-highest cluster entry ever after 2026-09-04 GPT-6 Astra 2080 pts; 4th-highest single-item of the entire 2026-08-07 streak). ~24 unique top+best items. **Row description refreshes on `/themes.md`** for the `MiniMax-m3` row (GPT-6 Sol and Luna 6th cluster entry) + the `hermes-agent-v0-19-quicksilver` row (Claude Code reads AGENTS.md only when telemetry is on [fixed] 374 pts + 212 cmts as the 2nd-highest-engagement AGENTS.md cross-walk standard validation signal of the 2026-08-07 streak, bug-report+fix complement to the 2026-09-19 `49760187` 702 pts primary signal) + the `agent-memory-architecture` row (Stripe's Knowledge AI Platform 122 pts + 75 cmts as direct cross-walk to the knowledge-base-as-agent-memory substrate pattern; The GitHub wiki is an anti-pattern 132 pts + 76 cmts as wiki-substrate pattern validation). Full delta list in the "Latest update — 2026-09-23 18:00 UTC" section on [`/sources/hackernews.md`](/sources/hackernews.md).

Prior run summary line for `git-repo-1 (2026-09-22T18:08Z)`: thirty-eighth pull of the 2026-08-07 streak; vault HEAD unchanged at `09bec367` for the 11th consecutive fetch; wiki HEAD bumped `d5a2b15` → `4ee7fc1` via 2 automated commits (`4e257c9` daily update 2026-09-22 + `4ee7fc1` sources-sync 2026-09-22 — thirty-ninth consecutive sources-sync); 8-dirty-file baseline; MOC-file modification persisting uncommitted for 9 calendar days; carry-forward bump 40/35/35/28 → 41/36/36/29-day. **No synthesis-layer changes**. Full delta list in the "Synthesis-policy — this run (2026-09-22T18:08Z)" section on [`/sources/git-repo.md`](/sources/git-repo.md).

Prior fetch (2026-09-22T18:30Z): web-search-1 attempted update **skipped** — TAVILY_API_KEY still empty, runtime registry reports connector as disabled, searxng-fallback slot empty. This is the thirty-ninth consecutive skip on the same code path; latest successful Tavily pull remains 2026-08-03 18:08 UTC. Full skip narrative in [`/sources/web-search.md`](/sources/web-search.md).

## Synthesis Themes (5)

The wiki tracks five recurring research themes, each documented as a row on [`/themes.md`](/themes.md) (with deeper evidence at `/themes/<NN>-<slug>.md`):

| # | Theme key | Status | Last seen | Strong evidence |
|---|-----------|--------|-----------|-----------------|
| 1 | `class-organoid-3d-bioprinting` | active | 2026-09-11 | 39+ evidence points across 2026-08-07 streak; Algolia + HN pulls; vault deep-reads |
| 2 | `MiniMax-m3` | watchlist | 2026-08-03 | open-weight-LLM cluster ~4,535+ pts across 15 items in 13 pulls over 16 days; unsloth/MiniMax-M3-GGUF local-LLM run path |
| 3 | `hermes-agent-v0-19-quicksilver` | active | 2026-08-03 | self-hosted-agentic-substrate cluster ~1,351+ pts across 8 entries in 15 days |
| 4 | `agent-memory-architecture` | active | 2026-08-31 | Almanac (YC S26) 3-pt adjacency + OpenWiki + Hermes + Obsidian-vault stack |
| 5 | `05-robin-research` | active | ongoing | personal research synthesis (organoid workflow + CelVivo/3Brain/Cellink equipment) |

Cluster totals updated for 2026-09-26: closed-weight-M3-comparison-anchor cluster **stays** at **~5,520+ pts across 6 items in 6 pulls over 9 days** (the **strongest sustained engagement cluster of the entire 2026-08-07 streak**, surpassing the open-weight-LLM cluster's 15 items / ~4,535+ pts) — the **6th PROMOTED entry** is the 2026-09-23 `49805509` GPT-6 Sol and Luna 1683 pts + 809 cmts. The 2026-09-26 `49849985` Revealing-the-details-of-how-OpenAI-agents-hacked-Hugging-Face 632 pts + 402 cmts is the **10th adjacent-not-promoted closed-weight-M3 cross-walk candidate** (would grow the cluster to ~6,152+ pts / 7 items but the entry is a swarmtraces.org security-incident writeup rather than a vendor release; adjacent-not-promoted per scope). open-weight-LLM cluster stays at **~4,535+ pts across 15 items in 13 pulls over 16 days** (the 2026-09-26 `49848269` Ollaya 553 + `49853175` Jev-wrapper 120 = **673 pts across 2 same-day items** are the **25th + 26th adjacent-not-promoted cross-walk candidates**); the **Jev-family cluster proliferation now totals ~3,626 pts across 8 Jev-family items in the 2026-09-09 → 2026-09-26 window** (System One Models + Jev 1728 + OpenJev 405 + Laya OS Jev Mac M4 CoreML + Kev Jev-like 317 + Jev-Leftpad 226 + Mini-AGI 227 + Ollaya 553 + Jev-wrapper 120 = **3,626 pts across 8 items in 18 days** — the **3rd-strongest emerging pattern** in the open-weight-LLM cluster's adjacency lane after the Qwen-3.8-family and the GLM-5.3-family). AI-vendor-business-stack cluster stays at **~4,104+ pts across 10 items in 8 pulls over 11 days** with the **OpenAI-agents-as-attacker sub-cluster at ~1,950 pts across 3 high-engagement items in 14 days** (2026-09-12 RubyGems-attack 889 + 2026-09-18 OpenAI-internal-repos-hack 429 + 2026-09-26 OpenAI-agents-HF-hacked 632) and the **AI-labs + math proof sub-cluster at ~2,621 pts across 7 items in 11 days** (2026-09-05 Anthropic FLT 706 + 2026-09-10 OpenAI proof-stolen 283 + OpenAI unpublished-math 210 + 2026-09-12 Dario Amodei 206 + 2026-09-24 Claude enzyme-system 747 + 2026-09-25 Anthropic-supply-chain-risk 196 + 2026-09-26 Terry Tao 273) — the 2026-09-26 `49849985` OpenAI-agents-HF 632 is the **25th adjacent-not-promoted AI-vendor-business-stack cluster entry candidate** + the 2026-09-26 `49852717` Terry Tao 273 is the **27th adjacent-not-promoted cluster entry candidate** + the 2026-09-26 `49856665` OpenAI bots meddled with US Government 16 pts is the **26th adjacent-not-promoted cluster entry candidate**. self-hosted-agentic-substrate cluster stays at ~1,351+ pts / 8 entries / 15 days (the 2026-09-26 `49854416` Floci 115 pts is the **9th adjacent-not-promoted cluster entry candidate**; the 115-pt engagement is below the cluster's typical 100-pt entry bar by a small margin but above the 21-pt Almanac refresh trigger). Tavily (web-search) remains the productive lane for vendor news; the Tavily skip streak continues until `TAVILY_API_KEY` is provisioned.

## Navigation Map

### Top-level entry points

- [`/quickstart.md`](/quickstart.md) — this page; orientation, themes, status.
- [`/themes.md`](/themes.md) — compact recurring-themes index; one row per topic key with First-seen, Last-seen, Confidence, Evidence count, Status, Evidence link.
- [`/open-questions.md`](/open-questions.md) — unresolved questions about the user's wiki / core memory model; sections Active / Answered / Stale.
- [`/commitments.md`](/commitments.md) — concrete work tasks, scheduled items, approvals, follow-ups with Owner when inferable.
- [`/personal-logistics.md`](/personal-logistics.md) — personal errands, appointments, pickups, travel, life-admin deadlines.

### Theme deep-dives

- [`/themes/01-organoid-equipment.md`](/themes/01-organoid-equipment.md) — organoid + 3D bioprinting equipment cluster.
- [`/themes/04-local-llm.md`](/themes/04-local-llm.md) — MiniMax-M3 / H3 local-LLM run path + open-weight-LLM cluster.
- [`/themes/03-ai-agents.md`](/themes/03-ai-agents.md) — Hermes Agent v0.19 Quicksilver release.
- [`/themes/05-robin-research.md`](/themes/05-robin-research.md) — agent memory architecture (OpenWiki + Obsidian vault + Hermes).
- [`/themes/05-robin-research.md`](/themes/05-robin-research.md) — Robin's personal research synthesis (organoid workflow + equipment + AI-agent integration).

### Source evidence

- [`/sources/README.md`](/sources/README.md) — connector status roll-up.
- [`/sources/git-repo.md`](/sources/git-repo.md) — Obsidian vault + OpenWiki wiki mtime + commits evidence (per source-specific instruction "Track vault + OpenWiki wiki mtime + commits (Robin 8-07 added)").
- [`/sources/hackernews.md`](/sources/hackernews.md) — HN Algolia + top/best feed pulls.
- [`/sources/web-search.md`](/sources/web-search.md) — Tavily / web-search pulls (currently 44-run skip streak; latest successful pull 2026-08-03).
- [`/sources/openwiki-version-drift.md`](/sources/openwiki-version-drift.md) — OpenWiki version drift / recovery notes.
- [`/sources/distributors-china-2026.md`](/sources/distributors-china-2026.md) — Robin 8-07 added: 8 untracked regional agent files (`02-Clients/Agents/`).
- [`/sources/literature.md`](/sources/literature.md) — literature deep-reads evidence.
- [`/sources/nocodb.md`](/sources/nocodb.md), [`/sources/ragflow.md`](/sources/ragflow.md), [`/sources/seafile.md`](/sources/seafile.md), [`/sources/wikijs.md`](/sources/wikijs.md) — daily-driver stack tools (count-data sources-sync targets).

### Operational reference

- [`/sources/OpenWiki速查手册.md`](/sources/OpenWiki速查手册.md) — OpenWiki quick-reference cheatsheet (4 bridge skills + workflows).
- [`/README-ACTIVATION.md`](/README-ACTIVATION.md) — OpenWiki operations manual (activation status + 4 connectors + daily cron pattern as of 2026-07-27).

## Connector Status (2026-09-26T18:00Z)

| Connector | Last successful | Last attempt | State |
|-----------|-----------------|--------------|-------|
| `git-repo-1` | 2026-09-26T18:09Z | 2026-09-26T18:09Z | active (42 pulls since 2026-08-07; vault HEAD unchanged at `09bec367` for the 15th consecutive fetch; 0 new vault commits since 2026-09-11; carry-forward bump to 45/40/40/33-day; 11-dirty-file baseline; MOC-file modification persisting uncommitted for 13 calendar days) |
| `hackernews-1` | 2026-09-26T18:00Z | 2026-09-26T18:00Z | active (46 pulls since 2026-08-07; `49849985` Revealing-OpenAI-agents-hacked-Hugging-Face 632 pts as highest-engagement item of the 2026-09-26 pull + 7th-highest single-item of the 2026-08-07 streak + 10th adjacent-not-promoted closed-weight-M3-comparison-anchor cluster entry candidate + 25th adjacent-not-promoted AI-vendor-business-stack cluster entry candidate; `49848269` Ollaya – Ollama for open-source Jev-style decision models 553 pts as 25th adjacent-not-promoted open-weight-LLM cluster entry candidate; `49852717` Terry Tao "more mathematicians" 273 pts as 27th adjacent-not-promoted AI-vendor-business-stack cluster entry candidate; Jev-family cluster proliferation at 673 pts across 2 same-day items Ollaya + Jev-wrapper = strongest Jev-family concentration of the entire 2026-08-07 streak) |
| `web-search-1` | 2026-08-03T18:08Z | 2026-09-26T18:07Z (skipped) | disabled (TAVILY_API_KEY empty + runtime registry reports disabled; 44-run skip streak continues; 45th skip would apply if attempted this run) |
| `google-1` | n/a | n/a | not configured |
| `notion-1` | n/a | n/a | not configured |
| `slack-1` | n/a | n/a | not configured |
| `x-1` | n/a | n/a | not configured |

## Prior Fetches (selected)

The full incremental history is preserved at the top of `quickstart.md` rev (git log `git log --oneline quickstart.md`). Selected prior fetches for context:

- **2026-09-26T18:00Z (hackernews-1)** — forty-sixth pull; `substring-noise-only Algolia lane on 2 of 6 + clean zero on 4 of 6` pull shape (distinct from the 2026-09-25 + 2026-09-24 + 2026-09-23 + 2026-09-22 + 2026-09-21 + 2026-09-19 + 2026-09-15 substring-noise-on-3-of-6 + clean-zero-on-3-of-6 prior pulls); **no cluster entry promotion**; **same-day Jev-family cluster proliferation (2 items, 673 pts in 24h) + one highest-engagement top-of-pull OpenAI-agents-as-attacker security-incident writeup + one named-stack-adjacent AI-labs-math cross-walk + one named-stack-adjacent self-hosted-substrate entry + one named-stack-adjacent Claude-Code-skill cross-walk + one named-stack-adjacent Claude-Code-Excalidraw entry + one named-stack-adjacent frontier-model-evaluation essay + one OpenAI-business-stack news item**; `49849985` Revealing-the-details-of-how-OpenAI-agents-hacked-Hugging-Face 632 pts + 402 cmts as the **highest-engagement item of the 2026-09-26 pull** + 7th-highest single-item of the 2026-08-07 streak + would-be 25th AI-vendor-business-stack cluster entry candidate (adjacent-not-promoted per scope because OpenAI is closed-weight and not on Robin's named daily-driver stack + the entry is a swarmtraces.org security-incident writeup rather than a vendor release); `49848269` Ollaya – Ollama for open-source, Jev-style decision models 553 pts + 134 cmts as 2nd-highest single-item of the pull + would-be 25th open-weight-LLM cluster entry candidate (adjacent-not-promoted per scope pending Tavily verification of open-weight status + ollaya.dev vendor identity); `49852717` Terry Tao "We're gonna need a lot more mathematicians" 273 pts + 378 cmts as would-be 27th AI-vendor-business-stack cluster entry candidate (1.39 cmts-to-upvote ratio = highest of the pull; adjacent-not-promoted per scope); `49853175` A single function Jev-like wrapper for LLMs 120 pts + 39 cmts as would-be 26th open-weight-LLM cluster entry candidate; `49854416` Floci: Locally emulating any cloud service 115 pts + 25 cmts as named-stack-adjacent self-hosted-substrate cross-walk; `49857528` Show HN: A Claude Code skill to analyze your chess games 41 pts + 29 cmts as named-stack-adjacent Claude-Code-skill cross-walk; `49857729` Make Claude your assistant in excalidraw 20 pts + 7 cmts as named-stack-adjacent Claude-Code + Excalidraw integration cross-walk; `49856665` OpenAI bots meddled with multiple US Government agency sites 16 pts + 5 cmts as would-be 26th AI-vendor-business-stack cluster entry candidate; `49849820` Analyzing Frontier Model Progress with Prince of Persia 28 pts + 22 cmts as watchlist-only frontier-model-evaluation cross-walk; 14th consecutive clean-zero `CelVivo` pull (longest zero streak of the 2026-08-07 streak); 2nd clean-zero `3Brain` pull of the 2026-09 streak; 10th consecutive fully-zero `openwiki` pull; 12th fully-zero `deepagents` lane (27 of 28 pulls since 2026-08-27 fully-zero on `deepagents`); Jev-family cluster proliferation pattern now totals ~3,626 pts across 8 Jev-family items in 2026-09-09 → 2026-09-26 window = 3rd-strongest emerging pattern in the open-weight-LLM cluster's adjacency lane; OpenAI-agents-as-attacker sub-cluster at ~1,950 pts across 3 high-engagement items in 14 days; AI-labs + math proof sub-cluster at ~2,621 pts across 7 items in 11 days. Row description refreshes on `MiniMax-m3` (Ollaya + Jev-wrapper + OpenAI-agents-HF + Terry Tao + Floci adjacencies) + `hermes-agent-v0-19-quicksilver` (Claude-Code-skill chess-analysis cross-walk) + `agent-memory-architecture` (Floci local-cloud-emulation cross-walk + Microservices-are-organizational-debt martinfowler.com MonolithFirst essay adjacency). Last seen on the MiniMax-m3 row stays at 2026-08-03 because the 2026-09-26 pull does not surface any MiniMax-M3 or MiniMax-H3 release.

- **2026-09-25T18:00Z (hackernews-1)** — forty-fifth pull; `substring-noise-only Algolia lane on 3 of 6 + clean zero on 3 of 6` pull shape (same shape as the 2026-09-24 + 2026-09-23 + 2026-09-22 + 2026-09-21 + 2026-09-19 + 2026-09-15 prior pulls); **no cluster entry promotion**; `49841563` Dutch government builds alternative for Microsoft based on NixOS 842 pts + 493 cmts as the **highest-engagement item of the 2026-09-25 pull** + the **8th-highest single-item of the entire 2026-08-07 streak** + a **named-stack-adjacent independent-substrate cross-walk candidate to the `agent-memory-architecture` row** (open-source-software-sovereignty framing + NixOS-deployment-target axis + structural analogy to OpenWiki independent-wiki-substrate pattern; adjacent-not-promoted per scope because the entry is a Dutch-government open-source-software-sovereignty story rather than an agent-memory canonical artifact); `49836374` Opus 5.5 is good at explainer videos 372 pts + 202 cmts as 8th adjacent-not-promoted closed-weight-M3-comparison-anchor cluster entry candidate (Anthropic Opus 5.5 capability writeup at launchvideo.io; 2nd Opus-5.5 capability writeup within 3 days after the 2026-09-22 `49803892` Anthropic Claude Opus 5.5 374 pts prior adjacent entry); `49845977` U.S. appeals court upholds designation of Anthropic as supply chain risk 196 pts + 256 cmts as 9th adjacent-not-promoted closed-weight-M3 cluster entry candidate + 23rd adjacent-not-promoted AI-vendor-business-stack cluster entry candidate (legal/policy news item rather than vendor release); `49844348` Scientists revive activity in frozen mouse brains for the first time 1 pt from Scientific American as brain-research adjacency to `class-organoid-3d-bioprinting` row's neuroscience-in-vitro application domain (NOT a true biological-organoid match; 1-pt engagement below the 100-pt refresh bar); 13th consecutive clean-zero `CelVivo` pull; 9th consecutive fully-zero `openwiki` pull; 11th fully-zero `deepagents` lane (26 of 27 pulls since 2026-08-27 fully-zero on `deepagents`); ~24 unique top+best items; `49845952` NSA Paying Billions to Test AI Models 118 pts + 62 cmts + `49844896` Microsoft Abandons Personal AI Chatbot Race with Copilot Reboot 52 pts + 38 cmts as 23rd / 24th adjacent-not-promoted AI-vendor-business-stack cluster entry candidates; `49837006` Show HN: Koi.rest 207 pts + 60 cmts as 3Brain substring-noise false positive (relaxation app, off-scope per scope). Row description refreshes on `MiniMax-m3` + `class-organoid-3d-bioprinting` + `agent-memory-architecture` rows.

- **2026-09-26T18:09Z (git-repo-1)** — forty-second pull; vault HEAD unchanged at `09bec367` for the 15th consecutive fetch after the Tomas 2026 + Vijakumaran 2026 resumption; wiki HEAD bumped `f9bb2e5e` → `4becee017` via 2 automated commits (`41347af` daily update 2026-09-26 + `4becee0` sources-sync 2026-09-26 — the **forty-third** consecutive sources-sync); 11-dirty-file baseline (down from 12-dirty at the 2026-09-25 fetch — the `41347af` daily update cleared `sources/README.md` from the dirty list); MOC-file modification persisting uncommitted for 13 calendar days; 45/40/40/33-day carry-forward bump (highest of the 2026-08-07 streak). **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-26T18:09Z)".
- **2026-09-25T18:17Z (git-repo-1)** — forty-first pull; vault HEAD unchanged at `09bec367` for the 14th consecutive fetch after the Tomas 2026 + Vijakumaran 2026 resumption; wiki HEAD bumped `17e9fafc` → `f9bb2e5e` via 2 automated commits (`069eef3` daily update 2026-09-25 + `f9bb2e5` sources-sync 2026-09-25 — the **forty-second** consecutive sources-sync); 11-dirty-file baseline (down from 12-dirty at the 2026-09-24 fetch — the `069eef3` daily update cleared `sources/git-repo.md` from the dirty list); MOC-file modification persisting uncommitted for 12 calendar days; 44/39/39/32-day carry-forward bump (highest of the 2026-08-07 streak). **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-25T18:17Z)".
- **2026-09-24T18:07Z (git-repo-1)** — fortieth pull; vault HEAD unchanged at `09bec367` for the 13th consecutive fetch after the Tomas 2026 + Vijakumaran 2026 resumption; wiki HEAD bumped `6aa1db2` → `17e9fafc` via 2 automated commits (`dc1a9fb` daily update 2026-09-24 + `17e9faf` sources-sync 2026-09-24 — the **forty-first** consecutive sources-sync); 12-dirty-file baseline (up from 10-dirty at the 2026-09-23 fetch); MOC-file modification persisting uncommitted for 11 calendar days; 43/38/38/31-day carry-forward bump (highest of the 2026-08-07 streak). **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-24T18:07Z)".
- **2026-09-26T18:07Z (web-search-1 attempted skipped)** — **forty-fourth consecutive skip**; TAVILY_API_KEY still empty + runtime registry still reports the connector as disabled; no Tavily request issued; no searxng-fallback result for the 2026-09-26 window. Latest successful Tavily pull remains 2026-08-03 18:08 UTC. See [`/sources/web-search.md`](/sources/web-search.md) "2026-09-26 attempted update — skipped".
- **2026-09-25T18:14Z (web-search-1 attempted skipped)** — **forty-third consecutive skip**; TAVILY_API_KEY still empty + runtime registry still reports the connector as disabled; no Tavily request issued; no searxng-fallback result for the 2026-09-25 window. Latest successful Tavily pull remains 2026-08-03 18:08 UTC. See [`/sources/web-search.md`](/sources/web-search.md) "2026-09-25 attempted update — skipped".
- **2026-09-23T18:00Z (hackernews-1)** — forty-third pull; `substring-noise-only Algolia lane on 3 of 6 + clean zero on 3 of 6` pull shape; **1 cluster entry promotion** — `49805509` GPT-6 Sol and Luna 1683 pts + 809 cmts → **6th PROMOTED entry in `closed-weight-M3-comparison-anchor` cluster**, growing the cluster from ~3,837+ pts / 5 items / 5 pulls / 8 days to ~5,520+ pts / 6 items / 6 pulls / 9 days (the **strongest sustained engagement cluster of the entire 2026-08-07 streak**); 11th consecutive clean-zero `CelVivo` pull; 7th consecutive fully-zero `openwiki` pull; 9th fully-zero `deepagents` lane; ~24 unique top+best items; `49806430` Pentagon AI-overreliance Iran 852 pts as 22nd AI-vendor-business-stack candidate; `49814947` Claude Code reads AGENTS.md telemetry-fixed 374 pts as 2nd-highest AGENTS.md cross-walk validation signal. Row description refreshes on `MiniMax-m3` + `hermes-agent-v0-19-quicksilver` + `agent-memory-architecture` rows.
- **2026-09-24T18:00Z (hackernews-1)** — forty-fourth pull; `substring-noise-only Algolia lane on 3 of 6 + clean zero on 3 of 6` pull shape (same shape as the 2026-09-23 + 2026-09-22 + 2026-09-21 + 2026-09-19 + 2026-09-15 prior pulls); **no cluster entry promotion**; `49820134` Claude discovers a novel enzyme system with CRISPR-like repeats 747 pts + 761 cmts as the **highest-engagement item of the 2026-09-24 pull** + the **5th-highest single-item of the entire 2026-08-07 streak** + **7th adjacent-not-promoted cross-walk candidate to `closed-weight-M3-comparison-anchor` cluster**; `49826221` Contrastive Language Models 148 pts + 46 cmts as 12th adjacent-not-promoted cross-walk candidate; `49833333` Are organoids useful for Alzheimer's research? 1 pt as **4th true biological-organoid Algolia title match of the 2026-08-07 streak** (same `abhishaike` owlposting.com author as the 2026-07-28 9-pt essay); `49833867` Show HN: Whiteboard (YC W26) 39 pts + 3 cmts as named-stack-adjacent `agent-memory-architecture` row cross-walk; 12th consecutive clean-zero `CelVivo` pull; 8th consecutive fully-zero `openwiki` pull; 10th fully-zero `deepagents` lane (25 of 26 pulls since 2026-08-27 are fully-zero on `deepagents`); ~26 unique top+best items. Row description refreshes on `MiniMax-m3` + `class-organoid-3d-bioprinting` + `agent-memory-architecture` rows.
- **2026-09-22T18:08Z (git-repo-1)** — thirty-eighth pull; vault HEAD unchanged at `09bec367` for the 11th consecutive fetch; wiki HEAD bumped `d5a2b15` → `4ee7fc1`; 8-dirty-file baseline; 41/36/36/29-day carry-forward bump. **No synthesis-layer changes**. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-22T18:08Z)".
- **2026-09-24T18:05Z (web-search-1 attempted skipped)** — **forty-second consecutive skip**; TAVILY_API_KEY still empty + runtime registry still reports the connector as disabled; no Tavily request issued; no searxng-fallback result for the 2026-09-24 window. Latest successful Tavily pull remains 2026-08-03 18:08 UTC. See [`/sources/web-search.md`](/sources/web-search.md) "2026-09-24 attempted update — skipped".
- **2026-09-22T18:30Z (web-search-1 attempted skipped)** — thirty-ninth consecutive skip; TAVILY_API_KEY empty. See [`/sources/web-search.md`](/sources/web-search.md) "2026-09-22 attempted update — skipped".
- **2026-09-21T18:17Z (git-repo-1)** — thirty-seventh pull; vault HEAD unchanged at `09bec367`; wiki HEAD bumped `95f286d` → `d5a2b15`; 10-dirty-file baseline; 40/35/35/28-day carry-forward bump. See [`/sources/git-repo.md`](/sources/git-repo.md) "Synthesis-policy — this run (2026-09-21T18:17Z)".
- **2026-09-20T18:30Z (web-search-1 attempted skipped)** — thirty-eighth consecutive skip; TAVILY_API_KEY empty.
- **2026-09-18T18:15Z (git-repo-1)** — thirty-fourth pull; vault HEAD unchanged at `09bec367` for the 7th consecutive fetch; wiki HEAD bumped `307e038` → `c6e81a5`; 11-dirty-file baseline; 37/32/32/25-day carry-forward bump.
- **2026-09-18T18:00Z (hackernews-1)** — thirty-eighth pull; `substring-noise-only Algolia lane on 2 of 6 + clean zero on 4 of 6` pull shape (highest single-day dedup count of the streak at 33 unique items + 10-overlap count tied for highest); 6th consecutive clean-zero `CelVivo` pull; 1st clean-zero `3Brain` pull after 18-consecutive-pull substring-noise streak; 4th consecutive fully-zero `deepagents` pull; 2nd consecutive fully-zero `openwiki` pull; **3 new would-be open-weight-LLM cluster entry candidates** + **3 new would-be AI-vendor-business-stack cluster entry candidates**, all adjacent-not-promoted.
- **2026-09-17T18:22Z (git-repo-1)** — thirty-third pull; vault HEAD unchanged; 36/31/31/24-day carry-forward bump.
- **2026-09-16T18:18Z (git-repo-1)** — thirty-second pull; vault HEAD unchanged for the 5th consecutive fetch; wiki HEAD bumped `5f33bd3` → `ac46732` via 6 automated wiki commits (3-day catch-up batch).
- **2026-09-16T18:00Z (hackernews-1)** — thirty-sixth pull; **9th `nbHits-positive-but-items-zero` materialization-gap pull of the streak** with the lowest 24-item dedup count of the streak; **first true biological-organoid Algolia positive hit of the 2026-08-07 streak** — `49729339` "Mice with human brain cells offer a tool to study disease" 2 pts + 1 cmt, NPR.
- **2026-09-11T18:14Z (git-repo-1)** — twenty-ninth pull; **vault HEAD bumped `391e4713` → `09bec367`** with **2 new vault commits** ending the 11-day vault silence streak: `1fd63019` 2026-09-10 (Tomas 2026 HGSC carboplatin/PARPi spheroid+organoid, Western University, 0 vendor callouts) + `09bec367` 2026-09-11 (Vijakumaran 2026 brain-organoid air-liquid interface oxygen engineering, first v4 Flywheel auto-sink pipeline entry).

## Backlog

Areas where evidence is insufficient or where future work is pending:

- **Tavily verification queue** — Adjacent-not-promoted cluster entry candidates need Tavily verification before promotion: `49752041` OpenJev 405 (open-weight-LLM cluster 19th candidate) + `49746618` Bonsai 2 27B 547 (20th candidate) + `49747925` Qwen 3.8 Omni Flash 313 (21st candidate) + `49752056` Microsoft AI scraping 753 (AI-vendor-business-stack cluster 13th candidate) + `49745940` Astra for Law 560 (14th candidate) + `49749656` OpenAI internal repos hack 429 (15th candidate) + `49717638` M4 Mac Mini Linux GPU Driver (local-LLM deployment adjacency) + `49753878` Harness Design for Coding Agents (Hermes Agent v0.20.0 implication) + `49724109` Kival self-hosted-knowledge-system (agent-memory-architecture adjacency) + the 2026-09-23 batch: `49806430` Pentagon AI-overreliance Iran 852 (22nd business-stack candidate) + `49812769` Jev in 25 Lines 513 (28th open-weight candidate) + `49817404` GPT-6 Astra drives 188 (10th closed-weight candidate) + `49813482` Tokens Too Cheap to Meter 137 (27th open-weight candidate) + `49817615` Gemini 3.8 TTS 118 (11th closed-weight candidate) + `49817289` Strands Harness 96 (10th self-hosted-agentic candidate) + the 2026-09-24 batch: `49820134` Claude discovers a novel enzyme system 747 (7th closed-weight-M3 candidate — would grow the cluster from ~5,520+ pts / 6 items / 6 pulls / 9 days to ~6,267+ pts / 7 items / 7 pulls / 10 days; scientific-discovery framing differs from the model-release framing of prior promoted entries) + `49826221` Contrastive Language Models 148 (12th closed-weight-M3 candidate) + `49833867` Show HN: Whiteboard (YC W26) 39 (named-stack-adjacent agent-memory cross-walk; 39-pt engagement well below the 100-pt refresh bar) + the 2026-09-25 batch: `49841563` Dutch government NixOS Microsoft alternative 842 (named-stack-adjacent independent-substrate cross-walk to `agent-memory-architecture` row via open-source-software-sovereignty framing + NixOS-deployment-target axis; would NOT promote the closed-weight-M3 cluster because the entry is an OS-substrate story rather than a model release) + `49836374` Opus 5.5 explainer videos 372 (8th closed-weight-M3 candidate — 2nd Opus-5.5 capability writeup within 3 days after the 2026-09-22 `49803892` Anthropic Claude Opus 5.5 374 pts prior adjacent entry; would grow the cluster from 6 promoted entries / ~5,520+ pts to 7 promoted entries / ~5,892+ pts but the entry is a third-party capability writeup rather than a vendor release) + `49845977` Anthropic supply-chain-risk appeals court 196 (9th closed-weight-M3 candidate + 23rd business-stack candidate — legal/policy news item rather than vendor release) + `49844348` Scientific American frozen mouse brains 1 (brain-research adjacency to `class-organoid-3d-bioprinting` row's neuroscience-in-vitro application domain; 1-pt engagement well below the 100-pt refresh bar) + `49845952` NSA Paying Billions to Test AI Models 118 (23rd business-stack candidate) + `49844896` Microsoft Abandons Personal AI Chatbot Race with Copilot Reboot 52 (24th business-stack candidate) + the 2026-09-26 batch: `49848269` Ollaya 553 (25th open-weight-LLM cluster entry candidate — open-weight status + ollaya.dev vendor identity pending Tavily verification; Ollama-for-Jev-style-decision-models cross-walk to the unsloth/MiniMax-M3-GGUF local-LLM run path) + `49853175` Jev-like wrapper blog 120 (26th open-weight-LLM cluster entry candidate — personal-blog Jev-wrapper essay rather than a vendor release) + `49849985` Revealing-OpenAI-agents-hacked-HF 632 (10th closed-weight-M3 candidate + 25th business-stack candidate — swarmtraces.org security-incident writeup rather than a vendor release; highest-engagement item of the 2026-09-26 pull + 7th-highest single-item of the 2026-08-07 streak; OpenAI-agents-as-attacker sub-cluster cross-walk) + `49852717` Terry Tao "more mathematicians" 273 (27th business-stack candidate — personal-blog mathematics essay rather than a vendor release; highest cmts-to-upvote ratio of the 2026-09-26 pull at 1.39 + highest single-item cmt count of the pull at 378; AI-labs + math proof sub-cluster 7th item) + `49856665` OpenAI bots meddled with US Government 16 (26th business-stack candidate — BBC news item rather than a vendor release; below the 100-pt refresh bar) + `49854416` Floci 115 (named-stack-adjacent self-hosted-substrate cross-walk to `agent-memory-architecture` row via local-cloud-emulation framing) + `49857528` Show HN: A Claude Code skill to analyze your chess games 41 (named-stack-adjacent Claude-Code-skill cross-walk to AGENTS.md + Codex CLI / Hermes Agent skill-file pattern; below the 100-pt refresh bar) + `49857729` Make Claude your assistant in excalidraw 20 (named-stack-adjacent Claude-Code + Excalidraw integration cross-walk; below the 100-pt refresh bar). All pending Tavily followup; the productive lane is web-search-1 once TAVILY_API_KEY is set.
- **web-search-1 unlock** — TAVILY_API_KEY needs to be set in the OpenWiki runtime environment + the runtime registry needs to refresh the connector's enabled state (the on-disk `config.json` shows `enabled: true` + 6 Chinese queries, but the runtime registry still reports the connector as disabled). 44-run skip streak continues until unlocked.
- **MOC-file working-tree modification** — `04-Literature/Deep-Reads/MOCs/MOC-类器官与3D培养.md` modification persisting uncommitted for 11 calendar days (mtime 2026-09-13 11:55:42 +0800); the file is being actively edited by Robin but the modification has NOT been promoted to a commit. Pending Robin's next vault commit cycle.
- **Carry-forward untracked vault entries** — `Review/zhou2023_computer_vision_microfluidics.md` 43-day carry-forward (one-off interpretation confirmed at 2026-08-28 fetch) + `celvivo_commercial/` + `templates/` 38-day carry-forward (both confirmed one-off alongside the `Review/` entry at the 2026-08-28 fetch) + `deepread_40240593/` 31-day carry-forward (mtime unchanged across 30 consecutive silent fetches since creation; working-folder pipeline hypothesis continuing to weaken). Synthesis layer unchanged because vault produced no new external-evidence commits.
- **8 untracked regional agent files from 2026-08-07** — `02-Clients/Agents/全国经销商地图-2026-0807.md` 1310 bytes + `02-Clients/Agents/华东-2026-0807.md` 10030 bytes + `02-Clients/Agents/华北-2026-0807.md` 13057 bytes + `02-Clients/Agents/华南-2026-0807.md` 10744 bytes + `02-Clients/Agents/山东-2026-0807.md` 8592 bytes + `02-Clients/Agents/湖北-2026-0807.md` 5766 bytes + `02-Clients/Agents/西北-2026-0807.md` 2287 bytes + `02-Clients/Agents/西南-2026-0807.md` 2657 bytes (all dated 2026-08-07 20:24:19 +0800). Persist unchanged in vault working tree. Documented in [`/sources/distributors-china-2026.md`](/sources/distributors-china-2026.md).