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

Agents log 2026-11-01

Author
Tom Rochette
History / Edit / PDF / EPUB / MD
Created: / Updated: / Status: finished / 33 minutes read (~6412 words)
agents 117
Table of Contents

(daily refresh, 1)
#

  • 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, five unchanged since their last checks (andyrewlee newest commit 2026-10-31 already resolved, ai-for-developers 2026-09-18, awesome-claude-code content 2026-10-22, e2b-dev 2026-08-21, slavakurilyak 2026-10-04), AMO additions 2026-10-31 to 11-01 resolved
  • cross-category: orchestration: the AMO #99 Herdr full-entry upgrade is source-side presentation of the herdr member, membership already holds, no action
  • cross-category: orchestration: joxo (JoxoAI/joxo, 89 stars) rejected on the thin-record tail, reopen on traction, coverage, or open-sourcing
  • section-root: agentic-coding-tools-landscape: rotation re-verification, DeepSeek Harness newest prerelease to dsh-v0.2.1-rc.5 (published 2026-10-30), ade-topic tail’s Superset renamed Paradev per its formerly-Superset README, Cursor pricing re-read with the X bundle unchanged, Gemini CLI docs stamp held at 2026-10-26, links re-fetched with Bloomberg and devops.com again 403.
  • section-root: the-agentic-development-environment-landscape: Superset renamed itself Paradev (ParadevAI org, Elastic License 2.0) and Paseo relicensed AGPL-3.0 to Apache-2.0 with Antigravity and Muse Code added, corrected in the leaderboard, top-five line, license axis, and references; all 17 URLs fetched 200 and Air’s claims re-verified unchanged.
  • section-root: model-selection-for-coding-tasks: Kimi K3 OpenRouter floors moved (input $0.149 on Wafer’s new fast row, output $7.15 on Sail Research’s fp4 row holding both legs at $0.1495/$7.15, Relace re-priced to $0.1824/$9.34, Makora to $0.50/$13.00, roster 30 to 29) with kimi-k2.7-code’s floor held at $0.60/$3.00; all seven provider pricing pages re-fetched with every other anchor unchanged; DeepSeek Search-and-Fetch billing reached its stated 2026-11-01 start with the page still carrying the free-through-October wording.
  • section-root: model-provider-feature-matrix: synced with the guide’s Kimi-floor moves (flagship cell split floors, workhorse cell held); all seven provider pricing pages re-fetched, every other cell unchanged.
  • section-root: context-management-patterns: link check, all nine cited pages fetched 200, no fact moved.
  • section-root: the-tells-are-structural: link check, all eight cited sources fetched 200, study figures unchanged, star figures untouched.
  • crewai: recorded 1.15.33 (2026-10-29) as the latest release on GitHub and PyPI.
  • crewai: replaced the Wikipedia reference with the deletion log and recorded the 2026-10-30 LLMPROD deletion of the article as a caution.
  • emdash: recorded v1.2.10 (2026-10-30) on the release train.
  • multica: recorded v0.7.2 (2026-10-30) as the latest release.
  • oh-my-codex: recorded v0.21.44 (2026-10-31) as the latest release, npm at 173 versions.
  • omnara: recorded omnarad v0.2.15 and cluster v0.2.15 (both 2026-10-30) on the two release lines.
  • sim: recorded v0.9.39 (2026-11-01) on the release train.
  • flowise: refreshed npm downloads to 11,853 in the month ending 2026-10-29.
  • ag2: corrected the stale PyPI reference description to the 1.2.2 tracking claim.
  • apowerb: corrected the stale registry reference description to version 0.2.97.
  • agent-teams: corrected the stale releases reference description to v2.18.4 latest.
  • agent-manager: recorded v0.46.0 (2026-10-31) as the latest release.
  • agent-swarm: recorded the v2.7.4 through v2.7.7 train (2026-10-30 to 2026-11-01).
  • agentlas-os: recorded v1.2.72 (2026-11-01) as the latest release.
  • crewplane: recorded v0.4.1 (2026-10-30) as the latest release.
  • lanes: recorded v0.50.0 (2026-10-31) as the latest release.
  • langgraph: recorded 1.2.19 (2026-10-30) as the latest release.
  • lobehub: recorded v2.3.0 stable (2026-11-01) on the release train.
  • looptroop: recorded v0.6.2 (2026-10-30) as the latest release.
  • openharness: recorded the desktop-and-web line at v1.3.44_desktop and v1.3.83_web (both 2026-10-30).
  • openrig: recorded v0.7.6 (2026-10-30) as the latest release.
  • orca: recorded v1.4.245 (2026-10-30) as the latest desktop release.
  • parallel-code: recorded v5.1.0 (2026-10-29) as the latest release.
  • paseo: recorded v0.14.0 (2026-10-29) stable and the v0.15.0-beta.1 prerelease (2026-11-01).
  • praisonai: recorded v4.7.22 (2026-10-30) as the latest release and praisonaiagents 1.7.20 on PyPI.
  • ruflo: recorded the v3.63.9 through v3.68.0 train (2026-10-29 to 2026-11-01).
  • t3code: recorded v0.1.4 through v0.1.6 as the stable line and the nightlies moving to v0.1.7.
  • tuios: recorded v0.12.1 (2026-10-31) as the latest release.
  • evoagentx: recorded the maintenance return continuing with pushes through 2026-10-31, release train still stopped at v0.1.4.
  • bifrost: release train unchanged past the 2026-10-29/30 components, repo pushed 2026-11-01.
  • experiential: pricing page plan-table rate rows cut to 10M/100M/1B tokens a day (from 9B/36B/54B) while the page’s own FAQ keeps the old ceilings, one-key model count 943 to 958, price history row appended; PyPI 0.7.225 at 228 releases, repo pushed 2026-11-01.
  • experiential: matrix billing-mechanics and price-trajectory cells moved with the 2026-11-01 rate-table cut.
  • litellm: train head moved to v1.109.0-dev.2 (2026-10-30), v1.106.6 still the latest stable and PyPI latest, repo pushed 2026-11-01.
  • llm-gateway: PAYG and DevPass pricing re-fetched unchanged, repo pushed 2026-11-01 with v1.23.0 still latest.
  • localai: repo pushed 2026-11-01 with v4.12.1 still latest.
  • ollama: repo pushed 2026-10-31 with v0.40.13 still latest.
  • nanogpt: rotation re-verification, every cited source unchanged, the $12/60M-token-week Pro tier stands.
  • cerebras-code: rotation re-verification, every cited source unchanged, Pro and Max still sold out at $50/$200.
  • requesty: rotation re-verification, the company’s own routing comparison rebuilt 2026-10-28 into Auto Routing content and now reading 600+ models, closing the 400-versus-600 disagreement, and a first HN story about the company at 1 point; the 5% markup unchanged.
  • synthetic: rotation re-verification, every cited source unchanged, Go $9 / Standard $30 / Big $90 tiers stand.
  • andrej-karpathy: rotation re-verified; Medium Software 2.0 page 403s automated fetches, reference now the 2026-10-06 archive snapshot; blog gained the HackerTube essay (2026-10-31); GitHub still ends at nanochat (2026-09-07).
  • armin-ronacher: rotation re-verified; newest essay Custom Voices With Qwen3-TTS (2026-10-29), thirteen essays July 4 to October 29; lucumr Algolia query still grounds the cited point counts (996 at 1,058, Tower at 558).
  • caleb-writes-code: rotation re-verified; newest upload CUDA is hard.. (2026-10-30), six uploads October 13 to 30; third-party FAQ reads 125K subscribers (2026-11-01), prose body still carries the stale record.
  • lars-faye: rotation re-verified; every cited URL live, feed still ends July 22, course price unchanged; no claim moved.
  • latent-space: rotation re-verified; newest issue Agents Prompt People (2026-10-31), AINews through October 31; www host now redirects to the bare domain and the AINews section moved to /news/; front page still shows no subscriber count.
  • ray-amjad: moved the Pricing section to its mandated position between Cautions and Compared to; no content change.
  • people-and-publications-feature-matrix: Ronacher cadence cell to thirteen essays July 4 to October 29, 2026; columns stay at thirty-three.
  • people-and-publications: link, front matter (no type field, mandatory tags), section-order, matrix membership/sort/width, and index-list checks clean across all 33 notes.

(daily refresh, 2)
#

  • a2a: re-verified, spec stays v1.0.1 and the official CLI train v0.3.1, the 11-01 push is docs-only (broken-link and partner-link fixes, an antitrust policy page).
  • acp: re-verified unchanged, trains still at v1.14.0 / schema-v1.28.0 and schema-v2.0.0-alpha.11 with the SDKs at Rust 3.6.0 and TypeScript 2.2.0, nothing newer tagged.
  • ag-ui: downloads refreshed to about 15.5M combined in the month ending 10-31 (core 10,196,314, client 5,307,098); a community Swift SDK for iOS and macOS merged 10-31 (PR #1512), not yet in the README SDK table; release train stays release/2026-10-30.
  • agents-md: rotation re-verified, every cited URL live, the 60k-project count and OpenAI’s 88 nested files stand, changelog entries and the mods directory unchanged; AAIF press reference moved to its canonical aaif.io/news address.
  • langchain-agent-protocol: rotation re-verified, every cited URL live, spec stays 0.1.6, bindings stay langchain-protocol 0.0.20, no dedicated launch thread, forks re-counted 63, push 10-30 dependabot-only.
  • uhp: re-verified, spec stays at its 2026-10-19 fifth version with the suite at 2026.10.19.post4 and the train at v0.37.13, the 11-01 push is CI and kit-rename housekeeping.
  • protocols-feature-matrix: AG-UI prose figure refreshed to about 15.5M combined downloads in the month ending 10-31; membership unchanged at thirteen.
  • protocols: matrix columns and category index list matched to the thirteen members in case-insensitive title order, all internal links live, no type fields.
  • chonkie: pushed 2026-10-31, trailing-month downloads re-verified unchanged, PyPI 1.7.0 remains latest.
  • code-graph-rag: PyPI train reached 0.4.5 (2026-10-30, 36 releases; the 0.4 line gates a repository’s .cgr.md out of the agent’s system prompt until approved), pushed 2026-11-01, 541 open issues.
  • knowhere: release v1.2.40 (2026-10-31, paired TOC boundary isolation recalibrated from the last confirmed physical page), pushed 2026-10-31; per-page pricing unchanged.
  • langchain: repository activity refreshed (17,103 commits, pushed 2026-10-31); pricing untouched.
  • ragflow: recorded v1.0.0-rc3 (2026-10-30, third and final release candidate per its own notes, rc2-schema compatible for direct upgrades); advisory record re-verified unchanged at eight.
  • open-codebase-index: pushed 2026-10-31 (out-of-rotation refresh on the work-list signal); npm train remains paused at 0.38.2.
  • semantic-code-search: rotation re-verification re-fetched every cited URL live; all claims unchanged (VS Code workspace-context page now dated 2026-10-28).
  • tree-sitter-chunking: rotation re-verification re-fetched every cited URL live; all claims unchanged.
  • retrieval-feature-matrix: maintenance row refreshed to 2026-11-01 (Code-Graph-RAG PyPI 0.4.5, Knowhere v1.2.40, RAGFlow v1.0.0-rc3); no membership change.
  • retrieval: links, front matter, and as-of dates verified across the sixteen members; matrix and category index page matched to membership (columns and entries sorted case-insensitively); category fit re-checked, no moves.
  • agent-sandbox: rotation re-verified, v1.0.9 (2026-10-29, router gRPC proxying and the router Helm chart) recorded with the v1.0.9 TCP 9090 NetworkPolicy widening added to the AllowAll caution
  • aicontainer: volatile refresh, npm trailing-month downloads 516 to 517 (dated-range window), stale 505 in Cautions corrected, pushes still renovate branches
  • artifact-fs: push advanced to 2026-11-01 on branch activity, main quiet since 2026-08-12, Artifacts pricing unchanged
  • brig: v0.5.0 (2026-10-31) recorded, breaking removals plus the v1alpha1 policy envelope, secret broker and renewer seams, Claude OAuth refresh
  • brood-box: rotation re-verified, v0.0.25 still latest, push on branch activity, HN search still zero hits
  • cubesandbox: rotation re-verified, post-release train through 2026-11-01, v0.7.3 still latest, HN threads hold at 7, 5, and 3 points
  • drop: rotation re-verified, firewall RuleSet train through 2026-10-30, v0.3.1 still latest
  • e2b: 2.62.0 (2026-10-28) recorded, PyPI at 2.62.0, pricing page re-checked unchanged
  • microsandbox: v0.7.12 (2026-10-31) recorded, nested virtualization support, cloud still private beta
  • nemoclaw: volatile refresh, exactly-one-prerelease story holds, open issues and PRs at 808
  • nono: v0.81.0 (2026-10-29) recorded and the fifteen-advisory disclosure record added (CVE-2026-47128 D-Bus sandbox escape, four high-severity bypasses), framed as disclosure evidence not an audit
  • nvx: daily prerelease train advanced to 2026-10-31
  • opensandbox: stable 1.2.0 (2026-10-27) recorded, OSEP-0025 egress work opened on main
  • openshell: v0.1.6 (2026-10-27) recorded, gateways now refuse unauthenticated startup, open issues back above the launch peak
  • overalls: trunk commits 1,746 to 2,243, launch still 3 November 2026
  • sandstorm: v0.6.0 release-prep pass recorded (glibc and Docker packaging repairs), no tag yet
  • yolobox: v0.19.12 and v0.19.13 recorded, train to seventy-two
  • sandboxing-feature-matrix: maturity cells refreshed for nine members, prose counts corrected to twenty-four, nono disclosure clause added
  • agentmeasure: v0.10.0 (2026-11-01) moves buyer review, task economics, period intake, cost drivers, and read-only MCP tools into a tagged CLI release; verovo PyPI package still unpublished (git-tag install), agentmeasure stays the last PyPI publish at 0.8.0
  • agentsview: token-usage docs flipped back to schema version 5 (version 6 documented 2026-10-26 through 10-31, the eighth recorded flip); no release after v0.46.1, only CI and dependency commits since
  • ccusage: no release after v20.0.46; trailing-month downloads refreshed to 813,384 (window 2026-10-03 to 11-01, ten zero days), claude-mem re-read at 72,879 in the same window
  • codeburn: unreleased fixes recorded (Cursor 3.27 safe-storage unwrap, PowerShell tool as shell tool, Capacity Dock provider persistence, Antigravity userMessage attribution); pushed 2026-10-31 with an 11-01 side-branch push, npm still 0.9.27
  • coucou: unreleased Windows and Linux merges from lazarevtill (Claude API model discovery, Flow save-failure survival) plus iPhone App Store iteration fixes since v0.5.8
  • token-monitor: v0.84.0 (2026-11-01) adds Manus credits, Muse web quota, the floating Edge Dock capsule, and plugin live rates; README coverage moves to 48-plus tools (Manus the newest account-level row)
  • git-ai: no release after v1.7.5 and no default-branch commit since 2026-10-20; the 10-31 push landed on side branches
  • tokscale: rotation re-verification; source table moved to 61 named clients (from about 50), trailing-month downloads 122,835 (window 2026-10-03 to 11-01), twelve unreleased fixes landed 2026-10-30, npm still 4.20.1
  • clawtrace: rotation re-verification, all cited URLs live, repository, plugin (0.1.23), pricing, and paper unchanged
  • agent-analytics: rotation re-verification, all cited URLs live, repository (pushed 2026-10-09, LICENSE still absent), npm CLI 0.5.37, and zero-hit HN footprint unchanged
  • session-analytics-feature-matrix: Tokscale cells and prose to 61 named clients, Token Monitor cells and prose to 48-plus tools, schema reading line back to version 5, breadth paragraph rewritten as a contested race (Tokscale 61 past agentsview’s roughly 60)
  • session-analytics: category index one-liners moved to Tokscale 61 named and Token Monitor 48-plus
  • agent-native: volatile refresh, npm stable held 0.3.55 with downloads at 7,277 over the pinned window 2026-10-25 to 10-31 and the pack at 25 skills, framework pushes remain monorepo app development
  • agent-plugins: rotation re-verification, all cited URLs live and every claim held, no figures moved
  • evoskill: rotation re-verification, all claims held, and the Status records the v1 paper’s venue, the First Workshop on Agent Skills (Agent Skills ‘26) at ACM CAIS 2026, per the v2 branch citation block

(daily refresh, 3)
#

  • software-factory: loki-mode - recorded the train passing v11.94.0 to reach v11.103.0 (nine minors since v11.94.0, npm in sync): .NET support with xUnit/NUnit/MSTest acceptance tests, package slimmed to 789 npm files with the Docker image now running the quality gates, run-branch cleanup with bundle saves and Bash secret-read denies, whole-issue runs up to four in parallel with splitting only after a failed whole attempt, labelled draft PRs for PARTIAL results, in-place answers to blocked runs with arbitrated Wall tests capping the Seal at PARTIAL; Docker pulls 306,269 (2026-11-01)
  • software-factory: fabro - volatile refresh: nightly train at v0.403.0-nightly.0 (2026-10-31), pushed 2026-11-01, nightly-only distribution unchanged
  • software-factory: fusion - volatile refresh: pushed 2026-11-01, v0.78.0-beta.9 and npm latest 0.77.0 unchanged with the stable lag at more than eight weeks, downloads re-baselined at 1,782 for the trailing month ending 2026-11-01, HN still no story
  • software-factory: machinist - volatile refresh: 179 commits past v0.4.0, main’s newest commit still 2026-10-29, v0.4.0 still the newest release, and the zero-coverage claim re-verified across three HN queries, all as of 2026-11-01
  • software-factory: superplane - volatile refresh: pushed 2026-11-01, v1.0.0 stable and fleet-manager v0.1.7 still the newest tags; no material change
  • software-factory: zeroshot - volatile refresh: train at v10.24.1 (2026-10-31, slow local Codex and Claude configuration probes tolerated up to sixty seconds), npm in sync
  • software-factory: fluent - rotation re-verification: every cited URL re-fetched (repository and README, the five release-notes documents v0.2.0 through v0.6.0, fluent.computer, the Discord invite, agentskills.io) and all claims confirmed, including v0.6.0 still the newest release and no push since 2026-10-11; the invite API no longer returns a member count
  • software-factory: super-simple-software-factory - rotation re-verification: every cited URL re-fetched (repository, example branch file tree with the .claude skills stamp, demo video, author site, huggingface/tau) and all claims confirmed, including the single commit on main and no push since 2026-08-04
  • software-factory: software-factory-feature-matrix - extended Loki Mode’s isolation cell for split-issue slices running in their own worktrees and branches (v11.99.0)
  • software-factory: : links, front matter, and as-of dates checked; matrix matched to membership and sorted case-insensitively; index entry list matched and alphabetical; category fit re-checked with no pivots
  • spec-driven-development: mcp-probe-kit - admitted from the Engineering4AI MCP Servers entry (merged 2026-11-01): the protocol-native spec workflow column, 41 stars, 859 npm downloads a month, zero coverage stated; note, matrix column, and index entry added.
  • spec-driven-development: cascade-v3 - rejected on the thin-tail precedent: spec-standard kind, 1 star, zero coverage, reopen on traction or coverage.
  • spec-driven-development: sdd-skill - rejected on the thin-tail precedent: spec-toolkit skill, 1 star, zero coverage.
  • spec-driven-development: design-os-studio - rejected from the category on the primary-artifact rule: the artifact is a declarative UI screen spec consumed by a design renderer, not a development contract; reported for surfaces.
  • spec-driven-development: intent-router - recorded the second deflation (239 and 51 on 2026-10-14 to 89 and 17 on 2026-11-01), the release list grown to ten entries with v1.4.0 still newest.
  • spec-driven-development: mycelium - recorded the v0.332.0 release (2026-10-31: native Agent Plugins 1.0 manifest, the gates run on Mistral Vibe for the first time).
  • spec-driven-development: openspec - recorded the v1.16.0 release (2026-10-30: breaking Node.js 22.12.0 floor for npm installs, standalone binary unaffected) and refreshed the download figure (3,243,520 a month).
  • spec-driven-development: pdd-kit - refreshed the volatile record: pushed 2026-10-31, npm downloads 200 to 1,101 a month.
  • spec-driven-development: arc-kit, bmad-method, gsd, spec-kitty, superpowers - volatile pass: push dates and download figures refreshed, no releases past the recorded ones.
  • spec-driven-development: cc-sdd - rotation re-verification (all sources alive): npm figure to 33,427 a month, release train unchanged at v3.1.0.
  • spec-driven-development: spec-driven-develop - rotation re-verification (all sources alive): repository still quiet since 2026-07-26, no facts changed.
  • spec-driven-development: spec-driven-development - category index entry list updated for MCP Probe Kit, OpenSpec one-liner’s download figure refreshed.
  • nimbalyst: recorded v0.82.4 (October 27, fixes across the file tree cap, external file opens, find-bar stalls, Claude Code sleep and question-card handling, Codex on Bedrock, and team tracker edits) atop the v0.82 pre-release train.
  • continue: repository metadata pushes continue into November with the read-only notice standing.
  • jetbrains: rotation re-verification; recorded Air Governance GA (October 22) and its $20/user/month Business plan, the results thread grown to 596 points, and the junie.jetbrains.com shell reading; the $8.33/$25.00 ladder stands dated 2026-10-22.
  • windsurf: rotation re-verification; both redirects re-confirmed by header and the Devin ladder re-observed unchanged on devin.ai/pricing (200 after the October 22 404), re-observed 2026-11-01.
  • px0: rotation re-verification; corrected the handoff list to the README’s seven harnesses (Gemini CLI dropped upstream), recorded multi-forge PR review (GitLab, Gitea/Forgejo, Bitbucket Cloud), corrected the v0.1.19/v0.1.20 release-notes reading, train paused at v0.1.20.
  • cursor: rotation re-verification; the November 12 OpenAI shutoff stands unrevised eleven days out and the X-fused pricing ladder is unchanged.
  • surfaces: volatile pass on the flagged members found no other movement (bb still desktop-v0.48.1, cc-haha still v0.7.6, CodeNomad stable still v1.0.1 with the dev train at v1.0.1-dev-20261101, OpenChamber still v2.5.3, puppyone still v0.4.0, Zed prices unchanged).
  • surfaces: links, front matter, as-of dates, matrix membership and case-insensitive sort, and the category index entry list all verified current; no matrix or index edits needed (no membership change, no cell-moving evidence).
  • task-management: agentrq - Recorded v0.15.2 (the agentrq skill served over MCP, agentrqd serving in the background by default) and refreshed counts.
  • task-management: beads - Refreshed counts (1,604 open issues, pushed 2026-11-01); v1.3.2 still the newest release.
  • task-management: kata - Refreshed counts (49 open issues and pull requests, pushed 2026-11-01); v0.20.0 still the newest release.
  • task-management: operon - Refreshed counts (28 forks, 17 open issues and pull requests); 4.4.0 still the newest release.
  • task-management: ordewell - Recorded v0.12.0 (the waiting-on-you inbox, task revert, notify command, minimal Claude Code tool sets, batched supervisor wakes) and refreshed counts.
  • task-management: pad - Recorded v0.19.0-rc.3 (account settings pinned to the signed-in account) and refreshed counts; v0.18.1 still the newest stable tag.
  • task-management: automaker - Rotation re-verification, repository unchanged, fork count corrected to 588; corrected the pricing claim, Anthropic’s article recording the 2026-06-15 Agent SDK credit-pool change paused before it took effect.
  • task-management: ccpm - Rotation re-verification, nothing material changed, fork count refreshed to 843.
  • task-management: plandb - Rotation re-verification, the stall holds, and the parent AgentField stack’s pricing refreshed to a published $0, one-tier statement.
  • task-management: task-management-feature-matrix - Refreshed the release and activity cells and corrected the Automaker Pricing cell.
  • margin-lab: volatile re-verification after the 2026-10-30 push (first since 2026-07-31) and the site’s model change: Codex tracker switched to GPT-6.1 Sol, new baseline from runs beginning 2026-10-30, eval-default commits by jjbouza, 30-day rates unchanged (81 percent Claude Code, 79 percent Codex), change panel now tags 1D/7D/30D statistically significant while the paused banner stands
  • artificial-analysis: rotation re-verification with every cited URL re-fetched: 718 models (from 703), Intelligence Index v5.0 (announced 2026-10-27, nine evaluations, Knowledge Work/Coding/Science/General at 25 percent each), Coding Agent Index v2.0, Gemini 4 Carbon new #1 at 59, changelog entries through 30 October, the three HN critical comments re-verified inside their items
  • benchlm: rotation re-verification with every cited URL re-fetched: 1,004 tracked, 724 benchmarks, 216 ranked (77 Supported), stamp October 29, leader Gemini 4 Carbon at 88.38 (Estimated, conditional range 78.7 to 98.1) with Claude Opus 5.5 at 86.29 the top Supported entry, confidence labels renamed High/Good/Moderate/Low
  • benchmark-radar: volatile re-verification: badge 41,974 records over 97 days as of 2026-11-01, pushed 2026-11-01, 152 open issues, only bot-recorded snapshots since the 10-30 merges, v0.13.0 still the newest release
  • bestvaluemodel: volatile re-verification: no changelog entries after the 10-31 DeepSeek price cut, bot refresh cadence continues, pushed 2026-11-01
  • trackers-and-leaderboards: matrix cells refreshed from the run’s re-verifications (Artificial Analysis 718 models, BenchLM 1,004/724/216/77 with the Oct 29 stamp, Benchmark Radar badge 41,974, Margin Lab Codex stack GPT-6.1 Sol/high); membership unchanged
  • trackers-and-leaderboards: category index one-liners re-dated (Artificial Analysis 718 models, BenchLM 1,004 tracked); links, front matter, and as-of dates verified across all 13 notes, matrix columns and index list matched to membership and sorted

(daily refresh, 4)
#

  • harnesses: ante - recorded v0.3.1 (October 30): ante auth list –json, session credential pinning, OSC 7501 turn status, Antix catalog moves to Opus 5.5, GPT-6.1 Sol, Kimi K3.1 Pro
  • harnesses: avante-nvim - recorded v0.5.0 (October 31), first feature release since v0.4.2: ACP support moved from Lua to a Rust library on the official ACP SDK, 0.6 to move the LLM backend; push refreshed
  • harnesses: codex - recorded rust-v0.166.0 stable (October 29): transcript selection with search, rendered Markdown Q&A, GPT-6.1 Terra across bundled/Vertex/Bedrock catalogs, codex queue –resume, doctor –network-only
  • harnesses: copilot-cli - prerelease train reached 1.0.106-1 and 1.0.106-2 (October 31 and November 1) behind the unchanged 1.0.105 stable; download window refreshed to 4,986,700 (month to November 1)
  • harnesses: fx - push as-of refreshed to 2026-11-01, v0.0.16 still newest
  • harnesses: goose - push refreshed to 2026-11-01, v1.57.0 still newest
  • harnesses: jcode - recorded the opt-in v1.0.0 preview line (previews .24 and .30, both October 28, experimental binaries directing production to the stable channel) behind the stable train parked at v0.99.1; pushes to November 1
  • harnesses: juggler - re-checked 2026-11-01, v0.7.9 still newest
  • harnesses: kilo-code - rotation re-verification, all references live: prerelease train to v7.8.25 and jetbrains v7.2.1-rc.1 (October 27-28) behind unchanged v7.8.18 and jetbrains/v7.1.12 stables; pricing page re-verified with Teams and Kilo Pass tiers unchanged
  • harnesses: mimo-code - rotation re-verification, all references live: commits 1,462, push 2026-10-31, v0.1.16 still newest, Token Plan and pay-as-you-go prices re-verified unchanged
  • harnesses: nanocoder - rotation re-verification, all references live: push 2026-11-01, npm downloads 6,649 (month to November 1), v1.31.0 still newest
  • harnesses: oh-my-pi - npm train reached 18.14.2 (October 31): withInactiveTools(context) breaking cleanup, intent-tracing blocking fix, steering-signal source carried to tools; push to November 1
  • harnesses: om-o - train reached v5.1.55 (November 1): programming-skill check-test-junk and mutation-proof Bun scripts; Go scaffolder LF fix in 5.1.54 (5.1.52-53 unpublished); push to November 1
  • harnesses: omnigent - push refreshed to 2026-11-01, v0.20.0 still newest
  • harnesses: onecli - recorded v2.17.0 (October 31, hours after yesterday’s rotation pass): PostgreSQL through the gateway, Zendesk and HeyReach integrations, readable approval cards
  • harnesses: opencode - v2 train reached v2.2.5 (October 30, five stables in three days): desktop worktree and shell-mode shortcuts, comment-popup quote sends; v1.x parked at v1.19.5; push to November 1
  • harnesses: openhands - push refreshed to 2026-11-01, v1.34.0 still newest
  • harnesses: openinterpreter - rotation re-verification, all references live: recorded rust-v0.0.57 (October 31, maintained Codex 0.159.0 compatibility profile validated); push to November 1
  • harnesses: qwen-code - recorded v0.26.5 (November 1): Generated UI default-on with removable card context, bounded SQL history queries for managed sessions, web-shell shortcuts; push to November 1
  • harnesses: unreal-agent - push refreshed to 2026-10-29 (as of 2026-11-01), v0.3.1 still newest
  • harnesses: zcode - rotation re-verification, all references live: desktop releases at v3.17.1 (October 30, read from GitHub releases after the site’s installer page stopped exposing versions to direct fetch), disclosure-thread point counts re-verified unchanged
  • harnesses: gemini-cli - rotation re-verification, all references live: commits about 6,569 (as of 2026-11-01), enterprise docs banner steering to Antigravity CLI confirmed, stable v0.66.0 unchanged
  • harnesses: claude-code - volatile check clean, no cited claim moved; no edit
  • harnesses: crush - volatile check clean, v0.99.0 still newest and nightlies continue; no edit
  • harnesses: junie - volatile check clean, 3975.1 (October 31) consistent with the note’s channel claims; no edit
  • harnesses: kimi-code - volatile check clean, kimi-cli archived with final push 2026-09-22 re-verified against the live API; no edit
  • harnesses: pi - volatile check clean, v1.4.0 still newest and the push-cadence claim holds; no edit
  • harnesses: warp-agent-cli - volatile check clean, no verifiable claim moved (docs page JS-rendered, repository stars frozen); no edit
  • baml: volatile re-verification after the 10-31 push, the 0.23.1 nightly train through the 20261030.a build (published 10-31), still no stable 0.23.1, forks 506 and open issues 366, npm trailing-month window re-read at 1,102,985 (2026-09-30 to 2026-10-29).
  • cua-s1: volatile re-verification, host pushed 2026-11-01, the cua-s1 component untouched since the 10-20 security fix, the 10-30 cua-spaces v0.19.0-v0.19.2 releases and the 11-01 driver nightly not touching it.
  • guardrails: volatile re-verification, pushed 2026-10-31 still branch activity, the default branch’s newest commit still the August hub-retirement merge, v0.11.0 still latest.
  • laya: volatile re-verification, 0.4.18 (10-30, the C++ ONNX backend laya_onnx), 0.4.19 (10-31, server-side routing diagnostics over HTTP), 0.4.20 (11-01, the Jev shared-state truncation warning) all published, PyPI at 0.4.20, pushed 2026-11-01.
  • tool-eval-bench: volatile re-verification, pushed 2026-11-01 on two CI-action bumps (#327, #328), still no release since v3.0.0, forks 55.
  • nanojev: rotation re-verification, all six references re-fetched and live, repo unchanged since 10-16 (forks 265), HN still 2 points zero comments, HF rolling counters re-read with a fresh as-of date (weights 1,284/88, dataset 1,370/11), stale day count refreshed.
  • openai-structured-outputs: rotation re-verification, all six references re-fetched and live, the gpt-6-luna pricing example and the guide’s gpt-6-astra lead unchanged, no claims changed.
  • anthropic-structured-outputs: rotation re-verification, all five cited URLs re-fetched and live, the seventeen-model supported list and the tool-use token table unchanged.
  • atomic-agents: rotation re-verification, all five cited URLs re-fetched and live, pushed still 2026-10-29 with the train and PyPI still at 2.15.0, nothing material moved.
  • instructor: rotation re-verification, all five cited URLs re-fetched and live, pushed still 2026-10-26 and PyPI still 1.17.0, nothing material moved.
  • hybrid-execution: matrix maintenance cells moved for the four push-signal members (BAML, Guardrails, Laya, tool-eval-bench), front matter to 2026-11-01.
  • hybrid-execution: link, front matter, section-order, chart-markup, and day-count checks clean across all 31 notes; matrix at 13 rows by 32 cells aligned; index entry list matched and sorted.
  • memory: claude-mem recorded v15.7.0 and v15.8.0 (My Work refresh no longer spills refused runs to the next logged-in agent after the hidden codex exec runs, newest-Haiku alias default, cloud viewer host and work_state sync); npm in sync at 15.8.0, trailing-month downloads 74,755.
  • memory: deja recorded the unreleased post-v0.23.3 train (recall edit-diff hunks and missing-term promotion, RRF fusion of vector-sidecar hits, shared index permissions, DEJA_HOME_ROOTS, handoff state); push 2026-11-01.
  • memory: gbrain train extended to v0.60.536.0 (quoted-as-source fences on every read, recall and list_pages source narrowing, opt-in private-page grants, GBRAIN_SESSION_AGENT, 3-page meaning-only search cap); pricing re-verified unchanged.
  • memory: hindsight pending train extended (decision model never writes observation text, evidence dates on every piece); no release after v1.1.0.
  • memory: openviking pending train extended with the Magpie integration plugin; no release after v0.5.2.1.
  • memory: plur recorded v0.23.0 (hash-chained history with plur verify, checkpoints and session manifests, opt-in experimental Swarm storage); npm in sync at 0.23.0; matrix audit cell moved.
  • memory: sno-station recorded v1.7.2 (conversations wait through login and usage interruptions) and the second release-history deletion (v1.7.1 gone from tags and releases); caution updated.
  • memory: rotation re-verification of claude-obsidian, memoryfields, metaclaw, zep, and file-based-agent-memory, every cited URL re-fetched live, no facts changed beyond zep’s Graphiti commit count (1,054 as of 2026-11-01).
  • memory: volatile re-verification of basic-memory, funes, memmachine, memu, and screenpipe found only routine pushes and a rolling CLI build tag, no note changes.
  • memory: TeleAI list’s late 2026-10-31 additions resolved: Engramory rejected below the 294-star floor, two Graph Memory paper-section entries rejected on the standing family.
  • evaluation-review: agent-qa - recorded the v0.2.4 (2026-10-31) and v0.2.5 (2026-11-01) releases and refreshed the activity line; npm tracks 0.2.5.
  • evaluation-review: crit - recorded the v0.23.2 release (2026-10-31) and refreshed the activity line and the releases reference.
  • evaluation-review: harnesstax - recorded another AgentBRANE dashboard-build republication (2026-10-30); the profiling traces remain unpublished.
  • evaluation-review: plannotator - recorded the v0.28.39 and v0.28.40 (2026-10-30) and v0.28.42 (2026-10-31) releases and refreshed the activity line.
  • evaluation-review: hermes-gate - activity line corrected: the 2026-10-29 repository push is a dependabot side-branch bump, the default branch last moved 2026-09-29 with a v0.1.8 branch prepared and unreleased; v0.1.7 unchanged.
  • evaluation-review: hunk - npm weekly installs refreshed to 8,157 (2026-10-23 to 10-29) and the activity line to the 2026-11-01 push; v0.24.1 unchanged.
  • evaluation-review: pushed-date refreshes folded into the deepeval, kiln, kitaru, langfuse, phoenix, and reviewbench activity lines, all releases unchanged.
  • evaluation-review: feature matrix maturity cells refreshed for agent-qa (v0.2.5) and Crit (v0.23.2), star figures untouched.
  • executions: aeon - refreshed counters (294 forks, 1,511 commits, pushed 2026-11-01) and proof-of-work counts (150 findings, 112 repositories, 5.4M stars)
  • executions: claude-code-action - release train to v1.0.266 (2026-10-31), counters 2,237 forks, 853 open issues and pull requests
  • executions: claude-code-hooks - fully re-verified in rotation with every cited URL re-fetched; the 33-event surface, five handler types, fail-open if filter, and the v2.1.267/v2.1.295 gates all re-confirmed; open issues roughly 16.2k
  • executions: claude-code-routines - fully re-verified in rotation with every cited URL re-fetched; added Ultra to the plan list and dropped the v2.1.211 gate, which left the live docs page
  • executions: codex-action - changelog extended with the 2026-10-30 UTF-8 subprocess-output commit; v1.13 still the newest tag; counters 182 forks, pushed 2026-10-31
  • executions: github-agentic-workflows - recorded the medium advisory GHSA-fqvc-336c-4fp9 published 2026-10-31 (OTLP credentials, ranges still unverified per the advisory body); corrected the 0.93-line prerelease clause; forks 614
  • executions: rivet-agents - refreshed counters (137 forks, 87 open issues and pull requests); pushed 2026-10-29 UTC unchanged
  • executions: n8n - re-verified unchanged ([email protected] a routine release, no note claim affected, forks still 61.3k)
  • executions: trigger-dev - recorded the v4.8.3 release (2026-10-30), current-version claim moved from v4.8.2; pushed 2026-10-31

(daily refresh, 5)
#

  • assistant-runtimes: - astrbot - recorded v4.29.4 (2026-10-31, portable message buttons, Decision Provider, SerpApi Google News search, bot-deletion cleanup, PyPI-mirror TLS verification), adoption numbers refreshed
  • assistant-runtimes: - anything-llm - rotation re-verification: every cited URL re-fetched live, v1.19.0 (2026-10-26) still latest, Cloud tiers (Basic $50, Pro $99) unchanged
  • assistant-runtimes: - nanoclaw - rotation re-verification: every cited URL re-fetched live, v2026.10.0 (2026-10-09) still stable with the v2026.10.1 line at rc.2, credential sentence corrected to the README’s pickable-gateway framing
  • assistant-runtimes: - qwenpaw - rotation re-verification: every cited URL re-fetched live, v2.2.2.post1 (2026-10-22) still newest stable, v2.3.0-beta.2 (2026-10-30, PyPI only) recorded behind beta.1
  • assistant-runtimes: - assistant-runtimes - volatile pass on the remaining ten flagged members: as-of figures and push dates refreshed, no release-line or status moves
  • assistant-runtimes: - accomplish-ai/coworker - rejected: dead tombstone repo (README-only “no longer supported”, no code, dormant since 2026-08-13), not the renamed fayazara/coworkers, which still exists unchanged
  • automated-research: aris - November 1 docs refresh recorded (bilingual News announcing the model worker, Goal Pulse reminder), following October 31’s optional model-worker MCP server for file authoring and writing skills, PR #486; release train still v0.4.28.
  • automated-research: kapso - main’s newest commits moved to the October 31 external-contributor packaging fixes (dev extra installs numpy and pandas, CI installs the dev extra alone); release train still v0.7.10.
  • automated-research: local-deep-research - November 1 batch recorded (Zotero sync result counts in the status table, provider usage-limit guidance preserved through research failures, parser memory-cap test); release train still v1.10.7.
  • automated-research: mole - six unreleased PRs merged October 31 recorded (evidence-based stopping on the traced headline figure, corrections and caveats extraction outside the claim cap, session-deadline call bounds, provider-refusal fail-fast, short-keyword lead searches, tunable verifier concurrency); v0.2.1 still latest.
  • automated-research: autosci - volatile check clean, the October 30 WeChat QR refresh was already recorded, nothing since.
  • automated-research: openresearch - volatile check clean, v0.2.23 still the latest stable release with nothing new since October 31.
  • automated-research: autocontext - rotation re-verification, every cited URL re-fetched and live, commit counter refreshed (2,144 to 2,172 of about 2,244), no material change.
  • automated-research: data-to-paper - rotation re-verification, every cited URL re-fetched and live, dormancy stands, no material change.
  • automated-research: gpt-researcher - rotation re-verification, every cited URL re-fetched and live, release and package lines hold, no material change.
  • automated-research: harmonic-aristotle - rotation re-verification, all seven citations live including the HN search re-run (6 hits) and the Erdos forum record of the Aristotle proof, no material change.
  • automated-research: math-inc-gauss - rotation re-verification, all five references live, the comparator table and the caught Codex and OpenCode workaround disclosures re-confirmed on the benchmark page, no material change.
  • automated-research: rotation pass done (autocontext, data-to-paper, gpt-researcher, harmonic-aristotle, math-inc-gauss); no membership, matrix, or index changes; full link sweep across all 251 category URLs found no dead links, bot-walled URLs (mathandai.org, erdosproblems.com, newscientist.com, venturebeat.com) confirmed alive on retry.
  • code-review: greptile - rotation re-verification: every cited URL re-fetched live, all prices unchanged, Enterprise forge row widened to Bitbucket Data Center, Gitea, and Perforce P4
  • code-review: ellipsis - rotation re-verification: every cited URL re-fetched live, all metered prices unchanged, the $100 credit now displayed on no fetched surface (org profile description emptied), CLI pushed October 31
  • code-review: kodus - volatile re-verification: push moved to October 31, no release after selfhosted-2.2.7
  • code-review: openqodex - volatile re-verification: push moved to November 1, [email protected] still latest, npm 1,276 weekly and 8,890 trailing month to November 1
  • code-review: qodo - volatile re-verification: push moved to November 1, v0.49.0 still latest
  • code-review: roborev - volatile re-verification: push moved to November 1, v0.74.0 still latest
  • code-review: feature matrix intro note count corrected from thirteen to fourteen, OpenQodex npm cell refreshed to 8,890 trailing-month downloads (2026-11-01)
  • code-review: links, front matter, as-of dates, matrix columns and row widths, and the index entry list all verified clean; chart markup compliant (theme=dark on dark sources only)
  • code-review: kodustech/awesome-ai-code-review unchanged (pushed_at still 2026-06-28), its scan carries over, q21 evidence re-confirmed
  • entroly: rotation re-verified; release 1.0.88 (2026-11-01) ships the PAC-KV compression, receipt-backed graph memory, and frontier codecs that main carried unreleased, plus 62 dogfood fixes and an action-gate scaffold; README documents Gemini CLI and OpenClaw hook surfaces; npm figure refreshed
  • graft: npm packages at v0.28.1 (2026-10-30, patch without a changelog section); growth numbers refreshed; stale old-package version corrected
  • rtk: dev-1.2.0 candidate train started (rc.599 as of 2026-11-01); stable remains v1.1.0; volatile numbers refreshed
  • jevgrep: pushed 2026-10-30 with no release; forks, issues, and npm trailing-month figure refreshed
  • headroom: pushed 2026-11-01 with no release; PyPI still 0.42.0; npm trailing-month figure refreshed
  • repomix: pushed 2026-11-01 with no release; npm trailing-month figure refreshed
  • repowise: pushed 2026-11-01 on main with no release after v0.59.0; forks, issues, npm, and PyPIstats refreshed
  • sol-pi: 2026-11-01 push resolved as non-default-branch activity again (zero main commits since 2026-10-26), non-material, note untouched
  • codealive: rotation re-verified, every cited URL intact, tier table and rates unchanged, no claim changed
  • sourcegraph-code-context: rotation re-verified, every cited URL intact, $16K floor and credits model unchanged, newest-blog-post claim re-dated, no claim changed
  • context-engines: matrix Entroly delivery cell updated for Gemini CLI and OpenClaw surfaces; membership and sort unchanged
  • context-engines: source Meirtz/Awesome-Context-Engineering unchanged since 2026-10-31 (zero commits), scan carried over, no new entries
  • mission-control: added from the task-management worker’s routing, the MIT self-hosted operations plane dispatching tasks to OpenClaw, Claude Code, and Codex with an Aegis review approval record gating completion (6,424 stars, alpha, pushed 2026-10-28)
  • control-planes: matrix extended to twenty-two columns with Mission Control in sorted position, index entry and Changes bullet added
  • agentfield: candidate line ran to v0.1.146-rc.10 with no new stable past v0.1.145, PyPI holding at 0.1.145
  • arifos: README known-gaps now record an external researcher security review in progress since 2026-08-25 with three findings landed; the self-measurement stamp held at 2026-10-29 (12 of 13 pass, F6 unmeasured)
  • cc-safety-net: npm 2.7.1 stops remote-installer interpreter and PowerShell bypasses and pins the recursive-delete boundary against working-directory arguments
  • hol-guard: coverage contract still unrenewed, 23 days past its 2026-10-09 expiry, page serving fetchers again, train holding at v3.58.0
  • openappa: release train to v0.38.0 (breaking authority-read change, 2026-10-31) and v0.38.1 (2026-11-01)
  • paperclip: backlog regrew again to 6,631, release line holding at v2026.1023.0
  • tracekit: release train through v1.5.1 and v1.5.2 (PyPI in step), both without release notes
  • veto: docs-only commit 2026-10-30, still no new tag past [email protected], Hosted tier selector first recorded
  • jamf-ai-governance: full rotation re-verification clean, no drift beyond the Browser Insights FAQ naming JNUC 2026
  • code-atelier-governance-sdk: full rotation re-verification clean, still dormant since 2026-07-23
  • control-planes: awesome-ai-governance unchanged since the 2026-10-31 check (newest commit the keel-verifier add), scan carried over

(daily refresh, 6)
#

  • surfaces: BBARIT rejected on the thin-record tail, binary distribution ended 2026-10-18, the repo is now a pointer to a source-only MIT drop at 13 stars, zero independent coverage under both current and former names
  • surfaces: Bodega One Code rejected on the thin-record tail, closed source with distribution repos at 47 and 2 stars, paid-tier licensing language withdrawn 2026-08-31, zero coverage
  • surfaces: DESIGN:OS Studio rejection carried onto surfaces by mechanism, the declarative screen-spec artifact fails the coding-support spine here too (spec-driven-development ruling of 2026-11-01)
  • orchestration: chorus admitted on the standalone-ADE line, AGPL-3.0 AI-DLC team harness, 1,216 stars as of 2026-11-01, 63 releases since June, launch-echo coverage (3-point Show HN), note, matrix column, and index entry added
  • orchestration: happier admitted on the OpenHarness multi-machine-command-center line, MIT E2E-encrypted control room for 100-plus agents, 2,102 stars, zero HN coverage stated
  • orchestration: nodeterm admitted on the terminal-family line, BUSL-1.1 Electron canvas terminal manager, 2,287 stars, 70 releases, $200/month Founding Sponsor tier seeded in price history
  • orchestration: agtx admitted on the terminal-family line, Apache-2.0 Rust blackboard ADE, 1,728 stars, train stopped at v1.0.6 (2026-09-16), cooling noted
  • orchestration: termio admitted on the terminal-family line, MIT macOS terminal-first ADE over the termiod daemon, 751 stars, v0.69.0 (2026-10-31), zero HN coverage stated
  • orchestration: cezar admitted on the Omnara/Paseo control-center line, MIT VPS cockpit with worktree races and review gates, 904 stars, launch-echo coverage (15 points)
  • orchestration: loushang (zhnt/loushang, 1,939 stars) rejected, its own agent kernel with no parallel/multi-agent/worktree surface, same-kind framework below the admitted tier while the tier question is open
  • orchestration: cavekit (JuliusBrussee/cavekit, 1,156 stars) rejected for orchestration, a compressed SDD plugin over one SPEC.md, frozen August 2026, family development lives in the caveman bundle held out-of-scope pending q9
  • orchestration: matrix extended to eighty-eight columns with the six topic-page admissions and the category index entry list extended, sorted, with Changes bullets
  • orchestration: kyrolabs/awesome-ade judged a credible dedicated mapped source for the category (25 stars, 18 member entries, actively pushed), wiring and its back-catalog walk proposed to the owner
  • hybrid-execution: portofcontext/pctx rejected on kind, a Code Mode tool-call transport and token-efficiency substrate with no structured-output, decision-model, or benchmark surface, the fourth and consistent decline after sandboxing, protocols, and executions
  • assistant-runtimes: accomplish-ai/coworker, the kaushikb11 list’s Openwork row, rejected as a dead tombstone (README-only, no code, dormant since 2026-08-13, 10,824 vestigial stars) and confirmed not the renamed fayazara/coworkers, which still exists unchanged

(daily refresh, 7)
#

  • orchestration: the ten remaining unlisted topic-page candidates rejected on the below-band line the six admissions set (termio’s 751-star zero-coverage floor): Pane 595, vicoa 523, wmux 440, agentbox 336 (identity-checked, not the madarco/agentbox member), mulmoterminal 256, operator-oss 208, rove 181, tlbx 116, agetor 110, cleancode 108
  • cross-category: get-bb/bb from the same sighting resolved as a membership match, the surfaces bb member’s own subject repository
  • cross-category: portofcontext/pctx closed out of scope with no claiming category after four kind declines, the Code-Mode tool-execution substrate question queued for the owner on the heretic precedent
  • orchestration: orchestration-feature-matrix counts, Changes bullet, and updated stamp brought level with the six-column extension the follow-up had left unstamped