↓ Skip to main content
tomrochette.com
  1. Agents/
  2. Agents log/

Agents log 2026-10-31

Author
Tom Rochette
History / Edit / PDF / EPUB / MD
Created: / Updated: / Status: finished / 18 minutes read (~3461 words)
agents 109
Table of Contents

(daily refresh, 1)
#

  • 1code: rotation re-verified, archive and March freeze stand, forks refreshed to 606
  • archon: recorded v0.13.1 and v0.13.2 (2026-10-29), the provider-plugin protocol 3 break and its same-day repair
  • dify: rotation re-verified, pricing unchanged at $590 and $1,590 per workspace per year
  • dmux: watch list holds at seventy-six days quiet, npm eased to 970 downloads for the month ending 2026-10-29
  • frink: rotation re-verified, no material change
  • clawteam: rotation re-verified, dormancy stands, no HN story exists for the project, lab attention on DeepCode (pushed 2026-10-31)
  • ivy-tendril: rotation re-verified, rewrite still quiet on both repositories, Pro still $59, docs hosts still failing DNS
  • swarm-orchestrator: recorded v15.0.0 (2026-10-30, verifier-first rebuild with swarm-verify 2.0.0), npm still at 14.4.0, downloads 540 a month, claims.md citation moved to main
  • agent-manager: recorded v0.45.0 (2026-10-30) as the latest release
  • agent-orchestrator: recorded v0.14.8 (2026-10-30) as the newest stable from the nightly train
  • agno: recorded v3.1.12 (2026-10-29, native approval flows for harness agents) on GitHub and PyPI
  • bernstein: rotation re-verified, HN search citation repaired to the query that returns both threads
  • bifrost: transports v2.4.0 landed 2026-10-29 adding Relays (non-LLM API calls through stored credentials with per-route pricing mapped to virtual-key budgets), telemetry v1.9.4 and semanticcache v1.7.4 on core v1.14.0, repo pushed 2026-10-31.
  • magnitude: @magnitudedev/cli 0.2.15 (2026-10-30) made the Anthropic-compatible messages and token-counting endpoints accept any anthropic-version header (or none), repo pushed 2026-10-30.
  • llm-gateway: rotation re-verification clean, PAYG and DevPass terms unchanged, DevPass catalog recorded at 100+ models, repo pushed 2026-10-31 with v1.23.0 still latest.
  • llm-gateway: DevPass catalog count corrected in the matrix models cell (100+ per the product pages, was 200+).
  • chatgpt-plans: rotation re-verification, plan prices unchanged ($8/$20/$100-$500, Business $25/user), GPT-6.1 Terra joined the rate table and the Work/Codex matrix (not Chat), GPT-6 Astra credit rates published (250 in / 1,250 out per 1M), GPT-5.6 Sol promo floor stated at 2026-11-21, GPT-5.5 retirement notice still displayed.
  • google-ai-plans: rotation re-verification, ladder, multipliers, storage, and country counts unchanged, dollar prices still rendering client-side, CA$13.99 Plus and CA$26.99 Pro unchanged, Gemini API promotional rates and the 2027 doubling unchanged.
  • minimax-coding-plan: rotation re-verification, M Plan prices, windows, and credit packs unchanged, M3.1 Flash Preview name retired with old-name requests served by MiniMax-M3.1-Flash, the token-plan FAQ URL now serves the M Plan FAQ.
  • model-access: litellm, LocalAI, and ollama releases (v1.106.6, v4.12.1, v0.40.13) all predate their 2026-10-29 verifications, no volatile work.
  • model-access: matrix matched to membership, columns sorted, rows at header width, index list matched and alphabetical, star-history markup correct in all nine repo notes, no broken internal links.
  • people-and-publications: lilian-weng - rotation re-verification: every cited URL live, homepage still ends at the 2026-07-04 harness post, all dated claims hold.
  • people-and-publications: ai-jason - rotation re-verification: latest upload still 2026-09-23 per RSS, third-party stats 234k -> 235k subscribers with 103 videos and 9.3M views unchanged; thoughtleaders page now bot-blocks plain fetches but serves through a browser client.
  • people-and-publications: birgitta-bockeler - rotation re-verification: series still 33 memos, newest still the 2026-10-13 colleague memo, TDD evaluation still her newest solo piece; fixed the stale sensors-article date in References (2026-05-07 -> 2026-05-27).
  • people-and-publications: steve-yegge - rotation re-verification moved the writing record: yegge.ai catalog 165 -> 170 essays, ~699k -> ~725k words, newest essay The Counting House (2026-10-27) returns to agentic architecture after the Gas Town shutdown lull; gastownhall/gastown pushed 2026-10-28 and not archived.
  • people-and-publications: deeplearning-ai-andrew-ng - Batch recency refreshed to issue 377 (2026-10-30, “Zombie Agents, Vampire Models, Witch Hunts, and the Mummy’s Curse: AI Fears for Halloween 2026”).
  • people-and-publications: matrix, category index, and membership verified in sync (33 notes, 33 columns sorted, every row at header width); no dead internal links, front matter clean (no type fields, all mandatory tags and audience_notes present); no entrant scan (no mapped sources for this category).
  • a2a: docs-only commit wave 2026-10-29/30 (partner-link fixes, LF antitrust policy), spec stays v1.0.1, CLI train stays v0.3.1, note stands
  • a2ui: tagged release train moved past the 2026-10-17 Python wave with the A2UI CLI’s first release cli-v0.1.0 (2026-10-30), a typesafe Python builder generator across protocol versions 0.9/0.9.1/1.0
  • acp: stable train moved to v1.13.0 with schema-v1.27.0 (2026-10-27) and v1.14.0 with schema-v1.28.0 (2026-10-30), v2 alpha train at schema-v2.0.0-alpha.11, Rust 3.6.0 (2026-10-28), TypeScript 2.2.0 (2026-10-30), agents list re-counted at 49
  • ag-ui: release train moved to release/2026-10-30, SDKs stay at 1.0.3, downloads refreshed (core 10,703,977 plus client 5,620,554 in the month ending 2026-10-30, about 16.3M combined)
  • agent-host-protocol: v1.0.0 remains the latest tag set, 73 open issues, pushed 2026-10-30, downloads refreshed (crate 345,702 lifetime, npm 590,346 for 2026-10-01 to 2026-10-30), the climbing-steeply npm reading corrected to a 590k-670k band since early October
  • uhp: release train through v0.37.13 (2026-10-31), conformance suite at 2026.10.19.post4 (2026-10-29) covering the harness_idempotency capability, spec unchanged at its 2026-10-19 fifth version
  • pic-standard: pushed 2026-10-28 on an external bounded refund sandbox verifier example (issue 198), spec train stays at v0.9.1 with 697 monthly PyPI downloads unchanged, 20 open issues
  • webmcp: draft report moved to its 2026-10-27 revision, the first API surface change since the declarative removal (cross-origin tool exposure under two-sided opt-in, section 6.3.4), 137 open issues
  • mcp: rotation re-verified, every cited URL live, specification revision re-confirmed at 2026-07-28 with the next revision still in draft
  • llms-txt: rotation re-verified, every cited URL live, spec site still v2 (modified 2026-08-10), Anthropic’s served file now 923 lines
  • anp: spec hub recovered after two unreachable runs and still serves ANP 1.2 with 1.0 and 1.1 archived, hub watch closed
  • protocols: matrix WebMCP maturity cell moved to the 2026-10-27 draft, AG-UI and AHP prose figures refreshed; membership unchanged at thirteen
  • code-graph-rag: refreshed repository activity (pushed 2026-10-31, 516 open issues); PyPI 0.3.13 unchanged
  • docling: PyPI 2.145.0 (2026-10-29) and repository activity (pushed 2026-10-31); trailing-month downloads unchanged
  • graphrag: refreshed repository activity (pushed 2026-10-29, 59 open issues); PyPI 3.3.0 unchanged
  • knowhere: recorded the v1.2.32-through-v1.2.39 release run (2026-10-27 to 2026-10-30), pushed 2026-10-30
  • langchain: refreshed repository activity (17,102 commits, pushed 2026-10-31)
  • lightrag: refreshed repository activity (pushed 2026-10-29, 427 open issues); PyPI 1.5.7 unchanged
  • llamaindex: refreshed open-issue count to 1,149 counting pull requests; push unchanged
  • open-codebase-index: pushed 2026-10-30; trailing-month installs read 5,569 to 2026-10-30, the decline string paused roughly flat
  • orama: rotation re-verification, every cited URL live; trailing-month range sum 6,041,061 to 2026-10-29 and last complete week 2,325,888, the prior weekly decline reversed inside npm zero-day noise; release stall unchanged
  • pageindex: SDK 0.5.0 (2026-10-29) and repository activity (pushed 2026-10-29, 153 open issues)
  • r2r: rotation re-verification, all dormancy claims hold (docs 404, sciphi.ai unreachable, PyPI 3.6.6, pushed 2025-11-07); open issues drifted to 132
  • ragflow: refreshed repository activity (pushed 2026-10-31); advisory record unchanged at eight
  • unstructured: PyPI 0.27.36 (2026-10-30) and repository activity (pushed 2026-10-30)
  • retrieval: matrix maintenance row refreshed to 2026-10-31; no membership change

(daily refresh, 2)
#

  • sandboxing: aigate - rotation re-verification: cited URLs all alive, push holds 2026-08-04, quiet window eighty-eight days.
  • sandboxing: clawk - rotation re-verification: cited URLs all alive, push holds 2026-08-13 at v0.4.0, quiet window seventy-nine days.
  • sandboxing: container-use - rotation re-verification: cited URLs all alive, release train still stops at v0.4.2, push holds 2026-09-21.
  • sandboxing: rotation pass done (aigate, clawk, container-use); no membership, matrix, or index changes; linked source unchanged, scan carried over.
  • agentmeasure: renamed Verovo 2026-10-30 (v0.9.0, “same measure, new name”) and pivoted to buyer-side charge and delivery verification, session checks demoted to Developer diagnostics; dead AgentMeasure Pages URLs repointed to Verovo, every cited URL re-fetched, counts refreshed (forks 10, open issues 9, pushed 10-31)
  • agent-causality: recorded v0.3.0 and plugin 0.1.7 (2026-10-30), PyPI caught up to 0.3.0 so the version-drift caution moved to a breaking-capture-contract caution, downloads refreshed (136 PyPI week to 10-31, 253 npm month to 10-29)
  • agentsview: volatile re-verification, no release after v0.46.1 with unreleased work continuing (Codex fast-tier pricing, Pi reasoning effort, OMP subagent parent links, Friction Log summary), docs still at schema version 6, counts refreshed (forks 671, issues and PRs 102)
  • ccusage: recorded v20.0.46 (2026-10-30), npm latest moved to 20.0.46, trailing-month downloads refreshed to 897,533 (window 09-30 to 10-29, seven zero days inside), forks 901
  • claude-code-agent-monitor: volatile re-verification, no release after v2.4.5 and no default-branch commit since 10-25 (pushes on side branches), open issues 50
  • claude-replay: volatile re-verification, the 10-31 push a second GitHub Actions bump, still no product commit since 09-18, trailing-month downloads 1,698 to 2,500
  • coucou: recorded Mac v0.5.4 through v0.5.8 (10-29 to 10-31, chat on your own AI subscriptions and Mail in the notch with search, auto-sort, and gated AI chat), Windows and Linux at 0.5.8 still unsigned, feature-gravity caution gained Mail, counts refreshed (forks 1,117, open issues 469)
  • novafabric: volatile re-verification, no release after v0.108.0 with dependency bumps and a DuckDB 2.0 fix unreleased 10-29, open issues 6 to 1, trailing-month downloads unchanged at 734
  • token-monitor: recorded v0.81.0 through v0.83.0 (10-29 to 10-31, experimental desktop plugins, quota alerts, official-catalog plugin updates, macOS floor to 13, network proxy, Max and Ultra naming), counts refreshed (forks 330, open issues 142)
  • ax: rotation re-verification, every cited URL re-fetched and live, release line and $20 team rollup unchanged, commercial-license contact now through Discord, counts refreshed (pushed 10-28, forks 14, open issues 42)
  • agents-observe: rotation re-verification, every cited URL re-fetched and live, repository state, release line, README, and launch thread unchanged, forks 71 to 75
  • codexmonitor: rotation re-verification corrected the release line to v0.7.67 (2026-03-24) from the v0.7.64 recorded since creation, dormancy unchanged, forks 415 to 417
  • session-analytics: matrix AgentMeasure Kind cell and reading prose record the Verovo rename and pivot, Token Monitor Deployment cell moved to macOS 13+, category index AgentMeasure one-liner notes the rename
  • agent-native: catalog grew to 25 skills (efficient-orchestrator added 10-28, efficient-frontier and efficient-fable now explicit-only aliases), npm crossed to 0.3.55 with pinned-window downloads at 8,514
  • anthropic-agent-skills: repo pushed again 10-28 (mcp-builder skill update to SDK v2, 61 commits), distribution-artifact reading held
  • gepa: volatile refresh, main pushed 10-30 (DSPy failed-output fix, contributor list, sampling fix), 144 open issues, PyPI still 0.1.4
  • sepia: volatile refresh, main unchanged since v0.12.4, pushed date moved to 10-30 on PR-branch activity, seven open issues
  • skillmd: CLI v1.3.4 shipped 10-28 (companion-gone-upstream install fix), index at 1,132,737 with the reviewed count holding at 1,947 and authors down to 30,538
  • skillopt: volatile refresh, two 10-29 security fixes (Sleep project-dir code import, web UI training-controls auth), 380 commits past v0.2.0, 62 open issues and PRs
  • skills-md: volatile refresh, pushes 10-28 to 10-30 are skill-execution fixes, npm CLI unchanged at 0.10.83, pricing unchanged at the repriced levels
  • skillspector: v2.13.2 shipped 10-28, a v2.13.1-branch patch backporting three false-incomplete-analysis fixes; open issues down to 147, main pushed 10-30
  • snyk-agent-scan: v0.6.10 shipped 10-30, formalizing the managed-mode hooks fixes and certifi bump as a release; PyPI at 0.6.10
  • agent-skills-open-standard: rotation re-verification, every cited URL re-fetched live, all claims held
  • opencode-skills-and-plugins: v2 became the shipping release train (v2.2.0-v2.2.5, 10-27 to 10-30, beside v1.19.5) and the ecosystem page was re-sectioned (31 plugins, 10 projects, 2 agents; the prior 43-plugin count covered all rows), rotation claims otherwise held
  • skills-feature-matrix: Agent-Native prose figures moved (25 skills, npm 0.3.55) and SkillMD as-of dates moved to 10-31; no cells moved
  • software-factory: fabro - nightly train at v0.400.0-nightly.0 (2026-10-28), still no stable tag.
  • software-factory: loki-mode - train at v11.94.0 (2026-10-31, npm in sync): measured Seal accuracy lines on signed receipts, resume for interrupted runs, loki try on public repos in a temporary clone, independent Java acceptance test, acceptance checks authored by a separate walled call; Docker pulls 296,635.
  • software-factory: machinist - 179 commits past v0.4.0 with main at 2026-10-29, still no release past v0.4.0.
  • software-factory: superplane - fleet-manager v0.1.7 (2026-10-28) advances the component train past the v1.0.0 platform tag.
  • software-factory: zeroshot - train at v10.24.0 (2026-10-31, npm in sync): contributed OpenCode and Hermes harness adapters beside the four core ones, hosted-run lesson reconciliation, README now accepts issues and contributions.
  • software-factory: software-factory-feature-matrix - Zeroshot cells extended for contributed adapters and hosted-run memory; awesome-software-factories 2026-10-29 sweep resolved (super-board below the bar).
  • ai-dlc: recorded stable v2.11.2 (2026-10-28, breaking for scripts), audit trail corrected to 131 events, largest HN thread now 29 points.
  • spec-kit: recorded v1.2.4 and v1.2.5 (2026-10-27/28), documented integrations moved to 38.
  • ai-unified-process: recorded the third release wave (2026-10-27/28, Vaadin v3.4.0, Spring Boot v3.4.1, Playwright v3.2.1, contract-first REST construction).
  • mycelium: recorded v0.331.0 (2026-10-28), the release gate now reads safety-skill assessment records.
  • spec-driven-development: matrix AI-DLC audit-trail cell corrected 113 to 131; source scan carried one addition (Nogging), rejected on the thin-tail precedent.
  • superset: recorded the repository rename to ParadevAI/Paradev (the superset-sh/superset path redirects, the README headlines “Paradev (formerly Superset)”), chart and references moved to the canonical path
  • orchestration: orchestration-feature-matrix reference for the Superset column moved to the Paradev canonical path
  • agentmeasure: recorded the 2026-10-30 rename to Verovo and the pivot to buyer-side charge and delivery verification in place; title and placement are owner questions
  • software-factory: super-board (EricTechPro/super-board, 564 stars) rejected on the Orbi coverage precedent
  • skills: thirteen VoltAgent-list entries rejected on the standing single-skill and pack exclusion, scan carried over

(daily refresh, 3)
#

  • surfaces: cc-haha - recorded v0.7.6 (2026-10-31, native Android and iPhone apps pairing over Cloudflare with the mobile UI Chinese-only, sleep prevention, in-chat scheduled-task management, edit-and-resend, an OpenRouter preset, and per-provider proxies) as the latest release; rotation re-verified, the HN search still returns zero stories
  • surfaces: matrix - cc-haha Mobile surface cell moved to native Android and iOS apps (v0.7.6, Chinese UI), VS Code + Copilot Scheduled work cell moved to the Agents window’s cloud automations (1.144, preview), cc-haha index one-liner mentions the mobile apps
  • surfaces: rotation pass done (trae, cc-haha, vibeyard, void); no membership, index-list, or sort changes
  • surfaces: cc-haha index one-liner gained its native mobile apps
  • cate: beta train at v2.1.0-beta.6 (October 29, the iOS agent-chat streaming fix, Cate Connect registration recovery, Chromium networking for local browser workspaces), pushed 2026-10-30
  • codeg: recorded v0.41.1 through v0.41.3 (October 28-30, folder-following tabs with split layouts and HTML previews, native MiniMax Code support and OpenCode 2 integration, smoother streaming replies, the Claude Code thinking-summary switch, and Codex follow-up prompt suggestions) as the latest releases
  • codenomad: recorded stable v1.0.1 (October 29, the twenty-mission project limit removed with per-mission storage, relevance-ranked history search, and the stale-transcript and Linux AppImage fixes) as the latest stable, the dev train re-tagged v1.0.1-dev
  • bb: recorded desktop-v0.48.1 (October 30, no written release notes) as the new latest stable
  • openchamber: recorded v2.5.3 (October 28, snippets with project and global scope, agent image generation through a picked image model, background auto-review that outlives the tab and stops after five rounds, and per-project header tabs) as the new latest release
  • whiteboard: recorded the monorepo’s train continuing past v0.4.0 under the diffr tag family (diffr/0.1.22 October 26, diffr/0.1.23 October 30, the model-and-effort pick for new whiteboards), repository pushed 2026-10-31
  • vscode-copilot: recorded the 1.144 release (October 28, Copilot Fusion generally available, cloud sessions and cloud automations preview, computer use experimental, assisted permissions generally available, Copilot Memory in harness sessions, image generation and connectors preview); Strengths harness list corrected for the 1.143 Local-harness retirement
  • trae: rotation re-verified, the $20/$60/$200 ladder, the free tier’s 5,000 completions and 2 cloud tasks, and the BytePlus enterprise path stand, trae-agent still stalled at its February 5 push with no tagged release
  • vibeyard: rotation re-verified, the quiet stands (pushed 2026-09-03, v0.3.8 still latest, the README’s coming-soon providers still unshipped, the one 2-point Show HN still the footprint)
  • void: rotation re-verified, the archive, the June 2 2026 last push, the v1.3.4 April 2025 freeze, the live beta-download product page, and both threads stand
  • agentrq: recorded v0.14.2 through v0.15.1 (2026-10-30 and 2026-10-31: key-file desktop secrets, throttle address-trust fixes, MCP server advertising, action-tool schema enums, agent-named kanban cards) and refreshed counts
  • epiq: recorded v1.16.2 (2026-10-28: follows a new comment on the watched ticket, Code tab commit-expansion fixes) and refreshed counts
  • operon: recorded 4.4.0 (2026-10-30: Status Dock Overdue and Today lists, occurrence-date filter and column, kanban presets grouping by independent properties, preset-specific resizable layout) and refreshed counts
  • ordewell: recorded v0.11.2 and v0.11.3 (2026-10-29: /resume alias, credentials guide, per-workspace Claude Code model pickers, mid-run repository config held for consent, completion-route shadow guard) and refreshed counts
  • pad: recorded the v0.19.0 release-candidate train (rc.1 2026-10-29, rc.2 2026-10-30: workspace-pinned-token behavior change, security updates, attachment previews, project catch-up, needs-you inbox, chat and push features)
  • backlog-md: refreshed counts (84 open, pushed 2026-10-29, 95,771 npm); v1.55.3 still newest
  • beads: refreshed counts (1,912 forks, 1,614 open issues, pushed 2026-10-31, 33,156 npm); v1.3.2 still newest
  • its-a-plan: refreshed counts (275 forks, 57 open, pushed 2026-10-30, 942 runner npm); v1.9.0 still newest
  • kata: refreshed counts (51 forks, 41 open, pushed 2026-10-31); v0.20.0 still newest
  • ai-dev-tasks: rotation re-verification with every cited URL re-fetched live; nothing material changed, still dormant
  • agent-kanban: rotation re-verification with every cited URL re-fetched live; nothing material changed, v1.13.4 still the newest tag
  • kanban-md: rotation re-verification with every cited URL re-fetched live; nothing material changed, v0.39.0 still the newest tag
  • task-management-feature-matrix: refreshed the release and activity cells for nine members after their notes’ volatile passes
  • task-management: dedicated-source check found no credible task-management-specific list; the broad sources keep carrying the category
  • task-management: q24 evidence refresh found no independent adoption or third-party coverage of TomzxCode/agentic; the rejection stands pending the owner’s ruling
  • benchmark-radar: volatile re-verification after the 2026-10-30 pushes: star CTA (PR #1027) and share-image serving onto the cards site (PR #1032) merged, badge to 41,457 records over 96 days as of 2026-10-31, pushed 2026-10-30, 144 open issues; v0.13.0 still newest.
  • bestvaluemodel: volatile re-verification after the 2026-10-31 push: eight-row add run on 10-29 halving Mistral Large 4 Preview, DeepSeek V4.1 Flash 1017 added 10-30 and repriced down about a third on 10-31, bot cadence unchanged, repository still unlicensed.
  • vectara-hallucination-leaderboard: rotation re-verification with every cited URL re-fetched: judge corrected to the commercial HHEM-2.3 per the README and model card, HHEM-2.1-Open the open variant, October 22 stamp and leader unchanged, downloads refreshed as a rolling window (~104k as of 2026-10-31).
  • llm-stats: rotation re-verification with every cited URL re-fetched: 408 canonical models, score v3.1, API tiers and quotas, 11 MCP tools, the two-person description, and both Show HNs unchanged.
  • trackers-and-leaderboards: feature matrix and index one-liner corrected for the Vectara HHEM-2.3 commercial judge; membership unchanged at thirteen.
  • trackers-and-leaderboards: dedicated-awesome-list check re-ran 2026-10-31 with no credible list; strongest new candidates 14 and 6 stars, both below bar.
  • section-root: model-provider-feature-matrix: Moonshot batch cell moved to 60% of standard on K3 and K3.1 Pro per the pricing page’s new batch statement, DeepSeek windows cell gained the Search and Fetch per-request billing starting 2026-11-01, and all seven provider pages re-fetched with the OpenAI cells holding (gpt-6.1-terra confirmed listed in the standard table at $1/$5, Astra at $10/$50 with the $1.00 cache read, gpt-5.6-terra and gpt-5.6-luna still listed)
  • section-root: model-provider-feature-matrix: Kimi floors moved on OpenRouter, the K3 floor to $0.1495/$7.15 on Sail Research’s fp4 endpoint with the aggregator listing holding at $0.192/$15.00, the kimi-k2.7-code floor to $0.60/$3.00 on CoreWeave’s new int4 row with Inceptron re-priced to $0.60/$3.50
  • section-root: model-selection-for-coding-tasks: full seven-page sweep recorded Moonshot’s 60% batch term and DeepSeek’s Search and Fetch billing clock (free through October, $0.75 per 1,000 requests from 00:00 Beijing time 2026-11-01), moved the Kimi floors, confirmed gpt-6.1-terra on OpenAI’s own pricing page, and re-dated the as-of clauses
  • section-root: context-management-patterns: rotation re-verification replaced three stale Claude Code documentation attributions, the retired “the context window is the most important resource to manage” quote, the removed “infinite exploration” anti-pattern, and the compaction-survival enumeration tightened to the current how-claude-code-works claim
  • section-root: the-tells-are-structural: rotation re-verification, all eight cited sources live with every figure intact, sepia and slop-grader star figures left untouched, as-of re-dated
  • cross-category: stars scan found zero new stars since the 2026-10-25 checkpoint, proposed checkpoint unchanged
  • cross-category: broad-source scan ran on all six sources, four unchanged since 2026-10-27, andyrewlee and AMO additions 2026-10-29 to 10-31 resolved
  • cross-category: orchestration: herdr-core (herdrdev/herdr-core, 191 stars) folded into the herdr note as ecosystem evidence, vendor-sibling mechanism
  • cross-category: orchestration: herduck (wenhanweime/herduck, 10 stars, self-described “Derived from Herdr”) folded into the herdr note as ecosystem evidence, derived-port mechanism
  • cross-category: alchemist-coder rejection carried by mechanism, the AMO-wave subject (Alekla0126/alchemist-coder) re-listed by andyrewlee 2026-10-31
  • cross-category: wsp, volt, borg, lookout, pohunek, yardlet, MercuryCLI, Aixle Flow, and sorcerer rejected on the thin-record tail, reopen on traction or coverage
  • cross-category: orchestration: Commonly (Team-Commonly/commonly, 1,374 stars) handed over on the OpenHarness standalone-ADE precedent, added to AMO’s Coordination And Team Systems 2026-10-30
  • cross-category: memory: andyrewlee added sno-ai/sno-station (the sno-station member) to its Multi-Agent Swarms section, membership already holds, no action
  • orchestration: commonly admitted as the category’s eighty-second member from the AMO coordination listing: Apache-2.0 team-pod workspace, 1,374 stars as of 2026-10-31, v3.0.0 (2026-10-28), 14,919 npm CLI downloads in October, zero HN coverage recorded as the signal; matrix column inserted, index entry added
  • herdr: recorded the ecosystem pair from the broad-source scan: the vendor’s herdr-core engine repository (193 stars) and the derived 10-star herduck terminal app
  • orchestration: nine thin-record tails held (wsp, volt, borg, lookout, pohunek, yardlet, MercuryCLI, Aixle Flow, sorcerer), alchemist-coder carried by mechanism, herdr-core and herduck folded into herdr
  • trackers-and-leaderboards: vectara-hallucination-leaderboard corrected to the HHEM-2.3-computed ranking with HHEM-2.1-Open as the open variant, note, three matrix cells, and index one-liner
  • surfaces: OpenBitFun rejection re-confirmed on the third q#14 re-read (3,423 stars, still a general-purpose workspace pitch)