Agents log 2026-11-13

(daily refresh, 1)

#

• harnesses: ante - recorded the v0.3.5 release (published November 12): Anthropic tool definitions and shared system prompt cached separately from the conversation, environment and memory-index details moved to turn start, Antix to Claude Fable 5.2, and empty-reply, resume-permission, and crash-safe-rewind fixes; rotation re-verification, all references live.
• harnesses: atomic-agent - rotation re-verification, all references live: push date refreshed to 2026-11-12, releases reference moved to v0.7.0 with desktop-v0.0.8, GAIA experiment page and product site unchanged, and the no-HN-thread claim re-checked (all hits still the unrelated Atomic Agents framework).
• harnesses: auto-dev - rotation re-verification, all references live: JetBrains download counters refreshed (Xiuper 700, Sketch 4,613) and the VS Code extension at about 5,547 installs, with repository metadata, the single-2026-09-25 commit cadence, the alpha-only release record, the org-to-personal redirect, and the Microsoft-name-collision HN record unchanged.
• harnesses: autohand-code-cli - recorded the v0.9.11 release and npm 0.9.11 (both published November 12): configurable subagent request limits with partial findings on exhaustion, ACP form elicitation, Code Mode image-evidence forwarding, and the Keenable search provider; rotation re-verification, all references live, pricing ladder re-read unchanged.
• harnesses: codex - recorded rust-v0.168.0 as the latest stable (November 13): a Sessions model-provider browser, an opt-in TUI workspace side conversation, audio-recording attachments, opt-in animated GIF generation, tmux passthrough image previews on Ghostty and Kitty, and OSC 777 notifications.
• harnesses: copilot-cli - recorded npm promoting 1.0.110 to stable (published November 12), which brought the masked git, ssh, and sudo password dialog out of the prerelease train as an /experimental opt-in with Entra sign-in, screen-reader, /usage wrapping, skill-picker, and MCP-config fixes, and refreshed the trailing-month download figure to 7,000,387 (window October 13 to November 12).
• harnesses: crush - recorded the v0.101.0 release (November 12): command-palette skill toggles with broken skills marked and explained, live per-tool-call timers, and request compression to Hyper.
• harnesses: deepseek-harness - recorded dsh-v0.2.2-rc.2 (published November 11: worktree review by branch, commit, and file type with direct GitHub review handling, named multi-layer SubAgent collaboration with a per-session AgentTeam, and a desktop update-channel switch) and dsh-v0.2.2-rc.3 (published November 13: an experimental DeepSeek chat import plugin and a Windows in-app update fix) taking the newest-tag spot; the project remains prerelease-only.
• harnesses: kilo-code - corrected the stable-train claim: v7.8.28 (November 7) is marked a prerelease on GitHub, so v7.8.27 and jetbrains/v7.1.12 remain the newest stables, with the JetBrains candidate train at v7.2.1-rc.3 (November 11) and the push date refreshed to 2026-11-13.
• harnesses: oh-my-pi - recorded npm reaching 18.20.0 (published November 12): 18.19.0 added the liquid-systemone judgment API for Liquid AI decision models and fixed Gemini CLI and Antigravity thought-signature ordering, and 18.20.0 added the google-decisions judgment API for Gemini’s Decisions API, moved prepareCompaction() onto the compaction.prepare loop phase, and surfaced apiKey on OAuthAccessResolution.
• harnesses: omnigent - recorded the v0.22.0 release (published November 13): automations triggering from Schedule, Jira, Slack, GitHub, Linear, or PagerDuty, chosen in the New automation form and startable from any session; push date refreshed to 2026-11-13.
• harnesses: openhands - recorded the v1.39.0 release (published November 11: automation dashboard owner filtering with a Service accounts option, Stop hook feedback kept out of chat, Paper theme printed as aged newsprint) and v1.40.0 (published November 12: improved Codex subscription authentication guidance and chat draft and history-page fixes); push date refreshed to 2026-11-13.
• harnesses: qwen-code - recorded v0.27.3 as the latest stable (published November 13: Web Shell composer chips for unfinished background work, Chinese localization for GitHub channel event prompts, and a 32-live-selection managed MCP cap), the v1.0.0-preview.1 cut (November 12), and the sdk-typescript-v0.1.33 and desktop-v0.27.3 tags; push date refreshed to 2026-11-13.
• harnesses: swe-agent - rotation re-verification, all references live: the maintenance-only banner, docs pages, both arXiv records, and the 307-point launch thread re-confirmed unchanged, the default branch quiet since 2026-07-16, the repository-level pushed date still moving on side branches (2026-11-09 the latest), and v1.1.0 still the newest release.
• harnesses: unreal-agent - rotation re-verification, all references live: the vendor blog, company site, the cited HarnessTax study, and the launch thread (247 points as of 2026-11-13) re-confirmed unchanged, the default branch quiet since 2026-10-06, and v0.3.1 still the newest release.
• harnesses: zerostack - recorded the v1.8.6 release (published November 11 on GitHub, crates.io still listing 1.8.5): deny_repeated_reads off by default, TUI span-normalization and ACP session-loading fixes, Pijul commands, XDG_DATA_HOME and API Route recipe docs, and a models.dev catalog refresh; crates.io counters refreshed to 3,799 total (643 recent) and the push date to 2026-11-11.
• harnesses: feature matrix verified matched to membership (45 columns, sorted, cell-count parity, Company-row alignment spot-check) and the category index entry list verified matched and alphabetical; front matter and internal links clean across all 45 notes.
• harnesses: kimi-code’s flagged signals verified with no edit (the note states no CLI version, so the 2.6.1 npm release touches nothing it claims); jcode’s v1 preview line advanced to preview.424 with no material change; exo, onecli, and warp release state unchanged against their notes.
• orchestration: agent-swarm - recorded v2.13.0 (2026-11-13) as the latest release.
• orchestration: agentlas-os - recorded v1.2.97 (2026-11-13) as the latest release, the push date refreshed to 2026-11-13.
• orchestration: agno - recorded v3.2.0 (2026-11-12) as the latest release on GitHub and PyPI with the A2A v1.0 breaking change summarized.
• orchestration: crewai - recorded 1.15.41 (2026-11-12) as the latest release on GitHub and PyPI.
• orchestration: foremerge - recorded v0.7.0 (2026-11-12) as the latest published release.
• orchestration: jean - recorded v1.0.15 (2026-11-12) as the latest release.
• orchestration: langgraph - recorded 1.2.21 (2026-11-12) on the core line and langgraph-cli 0.4.43 (2026-11-12).
• orchestration: lobehub - recorded v2.3.3 stable (2026-11-12) on the release train; docs root now redirects to the only-ai markdown path.
• orchestration: omnara - recorded omnarad v0.2.22 and cluster v0.2.22 (both 2026-11-13) at the head of the two release lines.
• orchestration: openharness - recorded v0.6.54_cli and v1.4.49_desktop (2026-11-13) on the two-line train.
• orchestration: openrig - recorded v0.7.13 (2026-11-12) as the latest release.
• orchestration: orca - recorded v1.4.258 (2026-11-13) as the latest desktop release.
• orchestration: pandacode - recorded the daily train reaching v0.3.123 (2026-11-13).
• orchestration: ruflo - recorded the v3.73 through v3.77 line (2026-11-09 to 2026-11-12) reaching v3.77.0.
• orchestration: sim - recorded the release train reaching v0.9.56 (2026-11-13).
• orchestration: autogen - last-pushed record moved to the 2026-11-12 docs-build commit; the README archive badge and caution re-fetched standing, and the GitHub repo now reads un-archived at the platform level.
• orchestration: 1code - the product domain stopped redirecting and now serves 503 errors; archive and March-dormancy record unchanged.
• orchestration: jetbrains-air - air.dev itself now redirects to www.jetbrains.com/air; the changelog resurfaced on the help site repurposed as a weekly Air Governance console changelog with no desktop version numbers; run-environments.html moved to air-teams-editor.html; 262.834.41 (2026-09-11) still the last verifiable desktop release.
• orchestration: multica - rotation re-verification, every cited URL live, no factual drift.
• orchestration: parallel-code - rotation re-verification, every cited URL live, no factual drift.
• orchestration: raven - rotation re-verification, every cited URL live, no factual drift.
• orchestration: scion - rotation re-verification, every cited URL live, no factual drift.
• orchestration: swarm-orchestrator - rotation re-verification, every cited URL live, no factual drift.
• orchestration: symphony - rotation re-verification, every cited URL live, the preview train still stopped in mid-September.
• orchestration: t3code - rotation re-verification, every cited URL live, no factual drift.
• orchestration: tuios - rotation re-verification, every cited URL live, no factual drift.
• orchestration: worktrunk - rotation re-verification, every cited URL live, no factual drift.
• orchestration: yylo - rotation re-verification, every cited URL live (site now canonicalizes to www.yylo.dev), the train still stopped at 0.2.1-rc.14 (2026-09-02).
• orchestration: orchestration - volatile pass refreshed push-date evidence across eighteen further flagged members (agent-manager, archon, autogpt, buzz, cezar, cli-agent-orchestrator, commonly, gascity, graphcode, happy-coder, multica, nodeterm, parallel-code, paseo, sandcastle, t3code, tuios, yao).
• orchestration: orchestration - matrix volatile alignment moved fourteen Current status cells to the newest release trains, star figures untouched.
• orchestration: orchestration - membership unchanged: index list and matrix columns verified matched at ninety entries each, license cells verified against every member note, zero front matter or link defects.
• bb: added community plugin directory to Status as ecosystem evidence (21 official plugins, retired marketplace, 25 npm bb-plugin packages)
• cate: recorded beta train reaching v2.1.0-beta.14 (November 12)
• cc-haha: recorded v0.7.8 (November 10, away-mode hardening, P2P direct, mid-turn permission switch)
• codeg: recorded v0.43.0 (November 12, conversation moves between folders, Web Service status-bar control, CI status on PR/MR lists)
• codenomad: recorded dev train re-tag to v1.1.0-dev (November 10), stable v1.0.1 unchanged
• nimbalyst: recorded stable v0.83.0 (November 10) after v0.82.8 graduated the pre-release train (Claude in Chrome, multi-project window moves)
• openchamber: recorded v2.7.1 (November 12, phone extensions, unified search, Memory organize)
• puppyone: recorded v0.4.3 and v0.4.4 (November 11, metadata-only notes)
• vscode-copilot: recorded the 1.146 release (November 11, Local harness deprecation with February 2027 removal, inbox mode, chat windows, Skills over MCP preview)
• whiteboard: recorded v0.5.2 (November 9, animation tours and clips)
• cursor: recorded the proposed November 12 shutoff passing with OpenAI’s temporary extension and the models docs unchanged
• delta: recorded the November blog run: Cloud Runners shipped (November 11), Claude Agent via ACP v2 (November 12), Conductor imports (November 10)
• vibeyard: rotation re-verified, quiet stands (train ends v0.3.8, about ten weeks)
• antigravity: rotation re-verified, models table and prices stand, November 18 removal dates five days out
• surface-feature-matrix: moved three cells on release evidence (Delta cloud, VS Code parallel, cc-haha mobile)
• surfaces: corrected the category index’s Cursor one-liner for the shutoff extension
• zed: pushed-signal liveness check only, no release tracked at note level, no edit
• openclaw: recorded the v2026.11.5 beta line advancing to beta.2 (2026-11-12)
• langbot: recorded v4.11.6 published and withdrawn the same day (concurrent token counts on a shared model rejected) and its replacement v4.11.7 (2026-11-11, the 4.11.x line’s final stable release, fixes only)
• openwork: recorded v0.18.67 (2026-11-13, AI Gateway features opened to every organization including self-hosted deployments), pricing re-checked unchanged (twenty-first-churn state holds)
• zeroclaw: recorded v0.9.0 (2026-11-12, the first release with non-operator sign-in: OIDC dashboard accounts, permission-scoped non-admin chat, a CLI-only server build, config schema v4)
• qwenpaw: recorded v2.3.0 going stable (2026-11-11 on GitHub and PyPI: Piccolo local decision models, Advisor Mode, experimental Python Code Mode, ReMe image memory, an OpenViking memory plugin)
• nanoclaw: recorded v2026.11.0-beta.2 (2026-11-12)
• picoclaw: rotation re-verification, every cited URL re-fetched live, v0.3.1 still the latest release, both README banners intact, default branch still quiet since 2026-08-19 with side-branch pushes through 2026-11-05
• nanobot: rotation re-verification, every cited URL re-fetched live, v0.4.1 still the latest release with Alpha current on PyPI, maintainer surname corrected to Cheng, release count corrected to eighteen, README recorded as no longer carrying the OpenClaw positioning
• assistant-runtimes-feature-matrix: volatile cells moved (LangBot v4.11.7, OpenWork v0.18.67, QwenPaw v2.3.0, ZeroClaw v0.9.0 plus its isolation cell, PrivateGPT pushed 2026-11-12)
• assistant-runtimes: q13 watch, openhuman reads 42,675 stars on 2026-11-13 against OpenClaw’s 394,415, the scale trigger stays unfired, openhuman held
• cross-category: stars scan empty, newest starred_at equals the checkpoint, agents/starred-checkpoint.txt rewritten with 2026-11-08T06:11:27Z
• cross-category: the andyrewlee 2026-11-12 batch resolves with no admissions: kraft 13 and mindflock 15 take the thin-record tail with orchestration kind flags, claude-bulkhead 34 the control-planes guard family, commonly a member sighting, the herdr-app/herdr-core split folds into the herdr note on the 2026-11-05 derivatives ruling
• cross-category: the AMO 2026-11-12 to 11-13 additions resolve with no admissions: kwant 557 holds below the 751 band on the amux coverage-quality reading, madsalt rejects as a services listing with no repository, scalekit’s second listing carries the 2026-11-05 enterprise-auth ruling, kraft a cross-list duplicate
• cross-category: yibie and handsome-rich pushes since their last scans are ticker and maintenance commits only (periodic evidence sweep, stale-entry refresh, weekly star counts), both q26 candidate lists carry
• cross-category: q26 evidence confirmed: yibie 795 pushed 11-13, handsome-rich 1,288 pushed 11-13, webfuse-com 2,591 stale since 09-21, WecoAI 1,077 stale since 07-30, AI4Scientist 188 pushed 10-28, lists stay unwired
• cross-category: q114 evidence unchanged, deep-research lists stay paper or prompts content (largest DavidZWZ 875 stale since 07-23), none above the bar
• cross-category: q118 liveness re-check, andyrewlee/awesome-system-one reads 59 stars pushed 2026-11-12, alive and actively curated, still unwired
• cross-category: q121 evidence refreshed, cuihuan/awesome-ai-gateway 156 stars pushed 11-12 (240-plus gateway catalog) and mahmoudsajjadi/awesome-ai-coding-tools 15 stars pushed 11-12, both far below the wired sources’ adoption floor, dedicated-source question stays with the owner
• cross-category: andyrewlee moved agent-swarm, agon, agent-kanban, and longhorizon-harness to Resting (#377); agent-swarm pushed 11-13 at 1,049 stars, agent-kanban pushed 11-13 at 502, longhorizon-harness pushed 11-11 at 1,871 (list classification contradicts live activity on all three), agon quiet since 10-06 at 57

(daily refresh, 2)

#

• agent-host-protocol: release train moved past 1.0, v1.1.0 (dated 2026-11-12) shipped 2026-11-13 as spec/v1.1.0 plus Go, Swift, npm, and crates all at 1.1.0, carrying the previously unreleased host-path semantics (#476), account selection (#583), MCP-served skills (SEP-2640), and experimental flow control (#492); default branch at PROTOCOL_VERSION 1.2.0, docs switcher at 1.2.0 draft plus 1.0.0 stable, 85 open issues and pull requests, crate 386,999 lifetime, npm 704,999 for 2026-10-14 to 2026-11-13
• webmcp: demand-side record re-grounded on implementation-status.md, ChatGPT Desktop supported since 2026-08-26, Edge 150 origin trial, Brave Leo AI default-on from Brave 1.102.2 (expected 2027-01-05, moved 2026-11-12, #383), Meta Ray-Ban Display coming soon, Mozilla position neutral and WebKit oppose; draft unchanged at its 2026-11-11 revision, 128 open issues as of 2026-11-13
• langchain-agent-protocol: rotation re-verification clean, every cited URL live, OpenAPI 0.1.6 and langchain-protocol 0.0.21 unchanged, pushed 2026-11-05, forks 64, footprint scans unchanged
• llms-txt: rotation re-verification clean, every cited URL live, spec site still v2 (modified 2026-08-10), Anthropic file 994 lines, footprint scan unchanged
• a2ui: pushed 2026-11-13 on docs-site restructuring only (docs/public removal and link retargeting), release train unchanged at cli-v0.3.0, spec hub unchanged at v1.0 current, no note edit
• acp: pushed 2026-11-13 on client-registry docs only (Sikemux, Yardlet, Tesseract clients added), trains unchanged at v1.17.0 with schema-v1.31.0, SDKs unchanged, no note edit
• uhp: default branch quiet since 2026-11-12 (v0.41.13), the 2026-11-13 pushed_at a side-branch push per the default-branch check, no note edit
• protocols: matrix AHP maturity cell moved to v1.1.0 and WebMCP criticism and adoption prose moved to the implementation-status record, AHP prose figures refreshed, no membership change
• context-engines: graphify releases v0.9.107 and v0.9.108 recorded (the MCP graph-reload stall fix, Oracle EDITIONABLE recovery, case-only node-id disambiguation, Gemini CLI extension), volatile counts refreshed, PyPI 0.9.108, hosted plans unchanged.
• context-engines: code-review-graph v2.4.0 shipped (GitHub release and PyPI 2.4.0 across 38 releases, replacing the not-yet-on-PyPI claim), unreleased 2026-11-12 fixes recorded, counts refreshed.
• context-engines: repowise main’s unreleased 2026-11-13 batch recorded (mcp.always_load, Redis streams, Temporal contracts, Kotlin binding), counts refreshed, PyPI 0.60.1.
• sol-pi: rotation re-verification clean, 2026-11-13 push again non-default-branch, counts refreshed.
• qmd: rotation re-verification clean, no push since 2026-10-15, v2.8.3 still latest.
• serena: rotation re-verification clean, trailing-month downloads 166,661 (was flat at 147,731), JetBrains installs about 20.4k, no claim changed.
• context7: 2026-11-12 to 11-13 default-branch wave resolved non-material (on-prem SSO and integration docs commits plus the @upstash/[email protected] patch of a package the note does not cover); note stands.
• context-engines: entroly, graft, and sol-pi pushes resolved as non-default-branch activity over quiet mains; notes stand.
• context-engines: jevgrep, headroom, rtk, toon-format, wrenai, and repomix pushes resolved as unreleased chores or features below the notes’ dated claims; notes stand.
• context-engines: link, front-matter, matrix (sort, 22-cell rows, per-column spot check), and index membership checks passed; no membership change.
• agent-native: volatile refresh, npm stable 0.4.4 to 0.4.6 (published 2026-11-12) with pinned downloads at 4,963 (11-07 to 11-13), pack held at 25 skills, framework pushes 11-12 to 11-13 still monorepo app development with two patch release trains
• agent-plugins: volatile refresh, main quiet since the 2026-11-06 v1.1.0 draft, the pushed_at move to 11-13 landing on side branches, v1.0.0 still the only tag, all claims held
• skillspector: volatile refresh, main pushed to 2026-11-13 (npm link slot pin fix) with open issues down 158 to 150, v2.13.4 held
• agent-skills-open-standard: rotation re-verification, every cited URL re-fetched live, all claims held a fifth time, the agent-table flags re-counted at 79
• skills-sh: rotation re-verification, every cited URL re-fetched live, leaderboard header 1,389,065 all time, design-mobile-apps to seventh as the window’s only High Risk, hyperframes-registry into the ten with Snyk Med, media-use Snyk to Med, feishu fleet Pending in rows 35-43 and 45-50, publisher chips split into multiple totals again, CLI to 3,435 forks and 676 commits (pushed 11-12), skillregistry.io still dark
• skills: SkillMD index one-liner moved to 2,204 of about 1.15M reviewed (was the stale 1,947 of about 1.13M)
• skills: three mapped sources unchanged since 2026-11-08 (no commits), scans carry over, no entrants
• docling: pushed 2026-11-13 with 992 open issues; PyPI 2.150.0 remains latest
• knowhere: pushed 2026-11-13 with 50 open issues unchanged; v1.2.55 remains latest
• lightrag: pushed 2026-11-13 with 419 open issues, aligned across sections; PyPI 1.5.7 remains latest
• pageindex: pushed 2026-11-13 with 172 open issues unchanged; PyPI 0.5.3 remains latest
• ragflow: pushed 2026-11-13 with 1,285 open issues; v1.0.0 remains latest and the advisory record re-checked unchanged at eight
• unstructured: pushed 2026-11-13 with 371 open issues; PyPI 0.27.37 remains latest
• semantic-code-search: rotation re-verification re-fetched every cited URL live, VS Code workspace-context page now dated 2026-11-11; all claims unchanged
• tree-sitter-chunking: rotation re-verification re-fetched every cited URL live; all claims unchanged
• retrieval: maintenance row refreshed to 2026-11-13 (PageIndex push date); no membership change
• memory: agentmemory - pending November 12 batch recorded (delete-path locking, session-cap eviction instead of refused writes, background keyword-index retries), downloads 86,955, push 11-13.
• memory: basic-memory - pending November 12-13 train recorded (Hermes PATH and shutdown fixes, structured MCP errors, portable ChatGPT upload), PyPI still 1.0.0, push 11-13.
• memory: claude-mem - v15.18.0-15.18.3 recorded (eve recording default-on with on-deployment redaction, worker startup and auto-update repairs, Windows Gemini hooks restored, Codex hook approvals persisted, opt-in local API token), npm in sync at 15.18.3, downloads 138,988, push 11-13.
• memory: cognee - v1.6.8 recorded (Slack connector and @cognee app, RabbitMQ embedding worker, Google-family connector hardening), PyPI in sync at 1.6.8, push 11-13.
• memory: deja - v0.24.2 recorded (turn-0 localization with verified guardrails, dead-end and FixPair recall, deja bench live and audit, native IntelliJ, Neovim, CodeCompanion, tmux extensions), README still 43 harnesses, matrix cell updated, push 11-13.
• memory: everos - pending 1.7.0 line recorded (breaking everalgo 2026-11 with per-task LLM and episode-derived profiles, online purge), GitHub and PyPI still v1.6.0, push 11-13.
• memory: funes - pending equal-score rank stabilization recorded, push 11-12.
• memory: gbrain - train continued to v0.60.708.0 (persistence/sync salvage wave, Hermes and pi merged pending release), push 11-13.
• memory: hindsight - pending 11-11-to-13 batch recorded (knowledge-bank search and multimodal queries, Gemma 4 retain fix, Hindsight app in OpenBot), push 11-13.
• memory: letta - pricing restructure recorded (API plan and per-agent metering gone, dollar credits, Pro includes $20 monthly credits), Price history row, release line v0.34.59.
• memory: memanto - pending Kimi Code integration and OKF YAML-date fix recorded, PyPI 0.2.31, push 11-13.
• memory: memu - pending Hermes hooks installer recorded, push 11-13.
• memory: openviking - python-sdk 0.1.20 and pending pgvector backend plus breaking Console change recorded, push 11-13.
• memory: screenpipe - pending app v2.7.147 batch recorded (focused-window capture ownership, enterprise simplified UI), push 11-13.
• memory: sno-station - README rework recorded, v2.0.1 still newest and still standing, push 11-13.
• memory: emulo - rotation re-verified, all URLs live, Emulo Proof v1 unexecuted runbook recorded.
• memory: engrim - rotation re-verified, all URLs live, no facts changed.
• memory: okf-agent-memory - rotation re-verified, all URLs live, no facts changed.
• memory: operator-memory - rotation re-verified, npm count corrected to 23 through 1.12.0 (codex 0.4.1), liao.gg serves a site-wide 403 bot wall, citation kept.
• memory: memory-feature-matrix - deja integration cell gained the v0.24.2 native extensions, Letta pricing cell moved to the credit-metered structure.
• memory: belief-context-graph - pushed_at movement resolved to side branches again, default branch quiet since 10-15.
• memory: memmachine - pushed_at movement resolved to side branches again, default branch quiet since 11-02.
• memory: plur - pushed_at movement resolved to further side branches, default branch quiet since 10-30.
• memory: zep - pushed_at movement resolved to side branches, default branch quiet since the 11-04 Community Edition removal.

(daily refresh, 3)

#

• aeon: counters refreshed (308 forks, 1,560 commits, pushed 2026-11-13), proof-of-work page unchanged
• airut: pushed 2026-11-12 recorded, counters otherwise unchanged
• centaur: counters refreshed (291 forks, 119 open issues and pull requests, pushed 2026-11-13), train still 0.1.213
• claude-code-action: train v1.0.275 (2026-11-12), counters refreshed (2,275 forks, 906 open issues and pull requests)
• claude-code-hooks: re-verified after v2.1.323, doc gates hold, open issues roughly 17.1k
• github-agentic-workflows: high advisory GHSA-4fwp-8v7m-jhhr (2026-11-11, patched 1.2.0) recorded, train v1.2.3 prerelease (2026-11-13), forks 626
• n8n: forks refreshed to 61,494
• rivet-agents: counters refreshed (140 forks, 94 open issues and pull requests), v0.5.6 still newest
• trigger-dev: v4.8.5 (2026-11-12) recorded, forks 1,535
• run-gemini-cli: fully re-verified in rotation, stalled-repo record holds (84 days), all claims confirmed
• agents-cron: fully re-verified in rotation, train 0.18.0, npm downloads 5,408 trailing month, coverage still empty
• executions: dedicated-list check found no credible source (ipanalytics/Awesome-Agent-Ops 1 star, dormant, not dedicated)
• atomic-agents: volatile re-verification after the 2026-11-13 pushes, three default-branch commits (#343 scraped-Markdown preservation, #365 arXiv archive names, #366 Wikipedia full-text fetch), no release past v2.15.0, pushed 2026-11-13.
• baml: volatile re-verification, pushed 2026-11-13, forks 518 and open issues 383, the 0.25.3 nightly train through the 20261110.a build (published 2026-11-11) with the stable train holding at 0.25.2, npm trailing-month re-read 1,387,816 (2026-10-13 to 11-11).
• cua-s1: volatile re-verification, host pushed 2026-11-13, the cua-s1 component untouched since the 2026-10-20 client security fix (commit path re-checked), the 11-12/11-13 release wave (cua-spaces v0.34.0-v0.36.5, spacesd v0.15.3-v0.16.1, sdk v0.25.2-v0.27.2, driver-rs v0.47.17-v0.49.1, bench v0.4.2-0.4.3, lume v0.7.1-0.7.2, sandbox v0.10.2) not touching it.
• instructor: volatile re-verification, pushed 2026-11-12 on branch activity with the default branch’s newest commits still the 2026-11-06 1.18-line work, PyPI 1.17.0, monthly downloads re-read 9,807,511, the rolling window up from the 8,516,567 that held flat for seven checks (no release, no trend read).
• valen: volatile re-verification, pushed 2026-11-12 on README and serving-command docs, no release past v0.4.0 and gguf-runtime-v0.2.0.
• valen: the Decision Index’s 2026-11-12 regeneration replaced the family’s three 1.0 rows with five Valen-Decision-1.5 rows (27B forty-seventh at 65.21, 9B 98th at 55.85, 4B 130th at 49.87, 2B 181st at 37.97, 0.8B 227th at 27.60), the 27B now above the vendor’s Jev 1.14 row.
• decision-index: board regenerated 2026-11-12 (301 configurations), karpo-decision-31B first at 73.86, GLiDE 27B second at 73.37, REAX 1.3 third at a re-read 72.64, new Perplexity Decider v1.3 rows (29th, 103rd), Valen’s five 1.5 rows entering, the vendor’s rows to sixty-seventh and seventy-sixth, the vision board at 85 entrants.
• jeeves: rotation re-verification, all five cited URLs live, repository unchanged since 2026-10-16, forks 24, weights re-read 124 downloads and 9 likes (window drift), thread 242 points.
• marvin: rotation re-verification, all eight cited URLs live, unchanged (pushed 2026-11-01 branch activity, PyPI 3.2.10, threads 127 and 20 points), forks 414, open issues 125.
• anyjev: rotation re-verification, all eight cited URLs live, repository unchanged (pushed 2026-10-21, v0.3.0 newest), mesa-anyjev still zero stars (pushed 2026-11-08), the family’s three board rows crowd-shifted to thirty-first, forty-second, and fifty-fifth at unchanged scores.
• clm: rotation re-verification, all six cited URLs live, repository unchanged (pushed 2026-10-05), checkpoint 1,092 downloads and 794 likes, both boards unchanged (JevBench 0.0 composite floor, Decision Index 6.49 at rank 293 of 301), Ollaya’s clm:8b row 0.357 unchanged.
• decider: Decision Index rows crowd-shifted at unchanged scores (31B 44th, pplx v1.2 26th, Flash v1.2 107th, chat 59th, second chat 106th, 35B 142nd, 4B 186th, 2B 240th of 301) with a new Perplexity v1.3 line (29th at 67.67, Flash v1.3 103rd at 55.14).
• h2o-lightning-4b: Decision Index sibling rows crowd-shifted (31B v1.3.0 16th, Bolt 74B 86th, E4B 138th of 301), the 4B still unmeasured on that board.
• jeff: the unattributed jeff-uncapped Decision Index row crowd-shifted to 294 of 301 at the unchanged 6.21; the family still carries no board row of its own.
• jev: the Decision Index’s vendor rows crowd-shifted to sixty-seventh (1.14, 60.24) and seventy-sixth (1.13, 59.22) of 301 at unchanged scores.
• kev: Kev 27B’s Decision Index row crowd-shifted to seventy-fourth of 301 at the unchanged 59.56.
• nimble: the Bespoke family’s Decision Index rows crowd-shifted at unchanged scores (v4.1 Max 15th, v3 124th, v2 165th of 301).
• strands-decider-2b: the Decision Index row crowd-shifted to 249 of 301 at the unchanged 20.72.
• von: the Decision Index row crowd-shifted to 282 of 301 at the unchanged 11.65; the archive holds.
• winnow: the Decision Index rows crowd-shifted at unchanged scores (12B 143rd, E4B 205th of 301).
• hybrid-execution: matrix maintenance cells moved for the five push-signal members and the Decision Index column; no membership change.
• hybrid-execution: category index one-liners for AnyJev and Decision Index moved to the 2026-11-12 board state; no membership change.
• git-lrc: volatile re-verification, pushed_at moved to November 12 on a side branch, main still October 23, v0.6.17 still latest.
• kodus: volatile re-verification recorded 2.2.14 (November 12, a Forgejo webhook hotfix), push November 13 with main active November 12.
• open-code-review: volatile re-verification recorded v1.13.8 (November 12: GitHub Copilot CLI integration, both-sides comparison review, MiniMax M3.1 Flash suggestions, task-template refactor, breaking --tools removal), 146 releases, push November 13.
• qodo: volatile re-verification, push moved to November 12, v0.50.0 still latest.
• roborev: volatile re-verification, push moved to November 13 with main active, v0.75.0 still latest.
• multi: rotation re-verification, every cited URL live; corrected the registry review-post record (removed 2026-10-22 through 10-30, restored by 11-13, conclusion unchanged), audit page live and still blocking auto-install, README gained Russian and Chinese translation links.
• graphite-diamond: rotation re-verification, every cited URL live, prices and page copy unchanged, both vendor announcement posts still 404, archived joins-Cursor capture live, HN threads unchanged.
• code-review: refreshed the feature matrix’s OpenCodeReview maturity cell to 146 releases (v1.13.8); the git-lrc, Kodus, Qodo, and roborev volatile re-checks found no cell changes, star figures untouched.
• evaluation-review: agent-qa - recorded the v0.3.5 release (2026-11-11, Sentry integration), npm tracks 0.3.5.
• evaluation-review: better-harness - recorded the PR 215 merge (2026-11-11, Qoder-session-detection analysis fix).
• evaluation-review: crit - activity line refreshed to the 2026-11-12 pinned-commit UI work, v0.23.2 still newest.
• evaluation-review: deepeval - recorded the Python 4.3.4 release (2026-11-11, PyPI-verified) and the 2026-11-12 ConversationalGEval fix.
• evaluation-review: harnesstax - recorded two more AgentBRANE dashboard-build republications (2026-11-12), traces still unpublished.
• evaluation-review: hermes-gate - activity line refreshed: the 2026-11-12 push is another dependabot bump, v0.1.8 still unreleased.
• evaluation-review: kiln - activity line refreshed to the 2026-11-12 and 11-13 default-branch merges, v1.2.1 still newest.
• evaluation-review: kitaru - activity line refreshed to the 2026-11-12 session-analyze and stale-docs merges, PyPI 0.27.7 still newest.
• evaluation-review: langfuse - recorded the v4.79.0 release (2026-11-12).
• evaluation-review: mlr-bench - rotation re-verification: every cited URL re-fetched, neurips.cc serving 500s site-wide with the acceptance grounded through the OpenReview record, the HF download figure refreshed (118 to 114, fresh as-of).
• evaluation-review: phoenix - recorded the arize-phoenix 21.1.0 release (2026-11-13), ending the platform’s hold at 21.0.0.
• evaluation-review: plannotator - recorded the v0.28.62 release (2026-11-12).
• evaluation-review: reviewbench - recorded the twenty-third reviewer onboarding (ncoder-sonnet-5-5, PR 113) and the leaderboard-configuration picker merge (PR 112), both 2026-11-12.
• evaluation-review: verbatim-citation-gate - rotation re-verification: every cited URL re-fetched and live, the third post-v0.3.1 gate fix recorded (2026-11-13, OpenRouter breaker panel), PyPI absence re-verified.
• evaluation-review: workshop - rotation re-verification: every cited URL re-fetched and live, pricing and the thin-footprint claims unchanged.
• evaluation-review: evaluation-review - matrix maturity cells refreshed (agent-qa v0.3.5, langfuse v4.79.0, phoenix v21.1.0), membership, sort order, cell counts, and column alignment verified unchanged.
• agent-sandbox: recorded v1.0.11 (2026-11-12, browser path-prefix routing with response rewriting, sandboxd header-token auth, SDK file-operation expansion, the PodTerminating readiness fix, AWS Nitro Enclaves and scoped-token MCP server examples, three breaking changes), releases to thirty-one, pushed 2026-11-12.
• cubesandbox: recorded stable v0.8.0 (2026-11-12, JuiceFS COW backend with cross-node resume, FS-only snapshots, restart policies, scheduler plugins with quality framework, E2B SDK 2.51.0+ compatibility over /v2/sandboxes, metrics and log collection, vsock to about 9.3 Gbps), pushed 2026-11-13.
• e2b: volatile refresh, post-2.63.0 train through 2026-11-12 CLI-only and unreleased, PyPI holds 2.63.0, pushed 2026-11-13.
• microsandbox: recorded v0.7.15 (2026-11-12, opt-in candidate installs and publishing, passthrough-mount operation cuts), pushed 2026-11-12.
• nemoclaw: volatile refresh, fix train through 2026-11-12 (workflow checkpoint recovery, TUI pairing, MCP timeout validation, installer OpenShell destination), prerelease-only release story holds, open issues 891, pushed 2026-11-13.
• nono: volatile refresh, train through 2026-11-13 (AWS GovCloud regions and S3 bucket-subdomain wildcards in the proxy, mise tool-versions fix), advisories re-checked at fifteen, pushed 2026-11-13.
• nvx: volatile refresh, daily prerelease train to 2026-11-13, pushed 2026-11-13.
• opensandbox: recorded stable 1.2.1 (2026-11-13, server, CLI, charts, images, and all five SDK packages published at 1.2.1; multi-version managed Firecracker bundles; Kotlin and C# recompile requirement), pushed 2026-11-13.
• openshell: recorded v0.1.8 (2026-11-10, auth_style credential-placement declarations, 403 on HTTP upgrades with explicit WebSocket rules, moved OCSF attribution fields, symlink-in-policy launch failures, SDK publishing moved to ai-openshell, gateway memory-leak and SigV4-token-log fixes), open issues 661, pushed 2026-11-13.
• yolobox: rotation re-verification, every cited URL alive, v0.19.13 remains newest, push holds 2026-11-01.
• sandstorm: rotation re-verification corrected the release record: v0.6.0 shipped 2026-11-01 (gVisor-default backend, Firecracker opt-in per node, one-binary master/node/console, mutual-TLS control plane, per-sandbox tokens, host groups, E2B SDK compatibility, three published fleet-measurement rounds), ending the five-month stall the note had recorded; open issues 14 as of 2026-11-13.
• overalls: out-of-rotation re-verification on the launch date passing: the banner still reads “launches on 12 November 2026” as of 2026-11-13 with no release or tag and the repository pushed 2026-11-09.
• sandboxing-feature-matrix: maturity cells refreshed for agent-sandbox, cubesandbox, microsandbox, opensandbox, openshell, and sandstorm, audit prose OpenShell parenthetical to v0.1.8, Sandstorm choosing bullet updated.
• sandboxing: index one-liners for Sandstorm and Overalls refreshed off stale release and launch-date claims.