
## (daily refresh, 1)

- anything-llm: recorded v1.20.0 (2026-11-17, browser extension rebuilt into a full sidebar chat with right-click add-context).
- astrbot: recorded the v4.30.0 line opening (alpha.1 2026-11-15, beta.1 2026-11-18, a migration release with backup-first upgrade warning).
- eigent: rotation re-verified; recorded v1.0.8 (2026-11-17, Home rail, global search, conversation timeline default); the pricing page's Free-card 1,000-registration-credits line returned (Price history row).
- hermes: rotation re-verified; recorded v0.23.2 (2026-11-16, about 1,989 PRs since v0.23.1, durable subagents disclosed in the release notes).
- nanoclaw: recorded the November beta line ending at v2026.11.0-rc.1 (2026-11-17, first release candidate, no more betas before stable).
- octop: rotation re-verified; recorded v1.0.3b6 (2026-11-15, Helm deployment, hosted experts, security fix) and v1.0.3b7 (2026-11-18).
- open-webui: recorded v0.13.4 (2026-11-17, Tether bring-your-own-computer headline).
- openclaw: recorded v2026.11.4 (2026-11-16) and v2026.11.5 stable (2026-11-17, restart-safe conversations) plus v2026.10.33, the first October-line extended-stable gateway patch (2026-11-18); Cautions issue count refreshed to 8.3k.
- openwork: recorded v0.18.68 (2026-11-18, build-split conventions); pricing churned a twenty-second time (add-ons returned, grandfather clause left, three main prices unchanged).
- assistant-runtimes: matrix Status row moved cells (AnythingLLM v1.20.0, AstrBot 1,248 issues with the v4.30.0 beta line open, Eigent v1.0.8, Hermes v0.23.2 and 54k issues, NanoClaw rc.1, Octop b7 and 680, Open WebUI v0.13.4, OpenClaw v2026.11.5 stable and 8.3k, OpenWork v0.18.68, PrivateGPT pushed 2026-11-18, ZeroClaw 1.3k).
- assistant-runtimes: trendshift.io now serves a 403 wall to scripted fetches; the octop badge claim grounded through the README's own badge markup (citation kept).
- nanobot: Cautions issue count matched to the Status line (868 -> 871) during the volatile pass.
- qwenpaw: Cautions issue count matched to the Status line (1,123 -> 1,162) during the volatile pass.
- zeroclaw: Cautions issue count matched to the Status line (1,246 -> 1,299) during the volatile pass.
- aris: default branch unchanged; the November 17 push was a side-branch creation (docs/community-qr-20261117), v0.4.28 still the latest release, no note edit
- autocontext: default branch unchanged since the November 7 milestone-20 merge; the November 18 push is side-branch and PR activity only, no note edit
- gpt-researcher: default branch unchanged since November 11; the November 12 push was a side branch, v3.8.0 still the latest release, no note edit
- autosci: default branch unchanged since the November 13 WeChat QR commit the note already records, no note edit
- autokernel: recorded the November 18 docs commit dropping the retired Forge enterprise link from the README, its URL now a generic contact page and the company site fronting a pay-as-you-go model API, with the pricing and vendor-caution lines moved and no tags or releases still true
- evo: recorded the November 17 Daytona SDK cap fix (issue 114), the main branch's first commit since October 5, with v0.8.0 still the latest tag and PyPI release and the fix unreleased there
- kapso: recorded the v0.9.4 release (November 17, on PyPI), which shipped the post-release commits the last refresh read in place (kapso report --json, the walkthrough video), beside kapso inbox --json, the knowledge-graph stores' unreachable-at-startup refusal, and the emoji-safe delegate TOML fix
- local-deep-research: recorded v2.0.1 shipping the same evening the last refresh ran (November 15, on GitHub, tags, and PyPI) with the prepared batch beside credential-scope changes (environment-set API keys bound to configured endpoints, REQUIRE_LOCAL flags enforced, backup-required migrations), and the November 16-18 post-release batch
- openresearch: recorded the November 16-18 post-release wave (settings tabs, OpenCode 2-only Auto mode, commit-scope chat titles, agent install diagnostics, steer retention, env-var request cards, chats without a project, the Working-and-Done inbox sidebar), v0.2.27 still the latest stable release
- harmonic-aristotle: rotation re-verification recorded the November 16 Rust expansion (unsafe-block verification, C-to-Rust equivalence, the Manifest and Project 0 Solana case studies, the Solana Foundation partnership), the November 12 multiclock and November 3 Ibex engineering posts, the DIII-D testimonial, and the 121-point Rust launch thread with its $500-free-credit cost report
- math-inc-gauss: rotation re-verification recorded the OpenGauss repository's removal from the math-inc org (a 0-star unlicensed Tkyle9439 mirror the only public trace) and the math.inc opengauss and formalqualbench pages' removal in the blog restructure, with the FormalQualBench repository surviving and status, pricing, and cautions moved
- openai-deep-research: rotation re-verification, all five references re-fetched and current with the same facts standing; the Wikipedia article is now nominated for merging into the ChatGPT article
- pion: rotation re-verification, all six references re-fetched and current, the launch thread still at 497 points; the Vending-Bench 2 leaderboard added Gemini 4 Hydrogen, Gemini 4 Argon, and GPT-6.1 Sol with Gemini 4 Carbon still leading
- agent-laboratory: rotation re-verification, every cited URL re-fetched and live, dormant since 2025-08-20 unchanged
- automated-research: matrix alignment verified (35 cells on every row, one Operator cell per member column against its note, Lean and Millennium rows positionally checked), all 35 files' internal links resolve, category index matched to membership and sorted
- code-review: coderabbit - rotation re-verification: every cited URL re-fetched live, all prices unchanged (Essentials $24, Team $48, Advanced $72 annual, $0.40 agent minute, $0.25 per reviewed file), homepage scale claims moved from 17K customers and 6M repositories to 20K and 8.5M.
- code-review: sourcery - rotation re-verification: every cited URL re-fetched live, prices and tiers unchanged, the unreleased 1.46.1 beta series grown from v1.46.1b7 to v1.46.1b54, PyPI and main still quiet since September 17.
- code-review: kodus - recorded the November 17 batch (2.2.15, web-1.2.14, selfhosted-2.2.11: implementation-check status decisions and paired eval cases); push November 18.
- code-review: multi - recorded the v1.22.0-v1.25.0 train (Devin CLI backend with SWE-2 models, agy read-only-gate bypass fixes, the permanent run journal, and a tagged-but-unpublished claude subscription backend); corrected the backend roster (Copilot no longer named), replaced the dropped one-model-refusal claim with the documented solo profile, and fixed the eval case count to 16 (the earlier line conflated the 12 graded known-bug pairs with the run's 16 cases).
- code-review: open-code-review - recorded v1.13.11 (branch-filtered session listing, Move review rules, path-spelling and multibyte fixes) and v1.13.12 (finding-identity fields kept separate, file_read byte bounds, non-zero partial-failure exits), 150 releases; push November 18.
- code-review: openqodex - recorded openqodex@0.19.0 (November 17: staged reviews gain second-file caller evidence, at most eight callers per declaration); npm reads 5,786 over the trailing month to November 18.
- code-review: qodo - volatile re-verification: push November 18 with v0.50.0 still latest; nothing else changed.
- code-review: roborev - volatile re-verification: push November 18 with v0.75.0 still latest; nothing else changed.
- code-review: code-review-feature-matrix - refreshed the multi (35 tags, eight-backend judge roster), OpenCodeReview (150 releases), and OpenQodex (5,786 npm trailing-month) cells; parity and sort order re-verified.
- code-review: the vendor-run kodustech list staleness re-confirmed for q21 and the dedicated-list search again surfaced nothing credible; no entrant candidates surfaced in scope this run.
- control-planes: agent-barn - volatile refresh: 20 open issues and pull requests, default branch on 2026-11-16 with unreleased workflow-approval work, the 2026-11-18 repo push a side branch.
- control-planes: agentfield - candidate line to v0.1.146-rc.38 (2026-11-16) with PyPI holding at 0.1.145.
- control-planes: arifos - counters refreshed (pushed 2026-11-16); the README verified-state block unchanged at its 2026-11-14 stamp.
- control-planes: cc-safety-net - npm train to 2.7.14 (Paranoid-only fix blocking a recursive `rm` piped through `xargs`) and 2.7.15 (container-isolation carve-out for read-only docker/podman/nerdctl runs).
- control-planes: cordum - default-branch push 2026-11-17 (dependency bumps, safety-evidence docs, integration-batch staging); no release past v1.3.1.
- control-planes: hol-guard - coverage-contract page still Cloudflare-blocked with the product page serving, 40 days past the assertions' expiry; counters refreshed.
- control-planes: ibm-contextforge - rotation re-verification clean; 994 open issues and pull requests, default-branch push 2026-11-17 with unreleased OAuth validation work for A2A agents.
- control-planes: jamf-ai-governance - rotation re-verification clean; no drift across all six references.
- control-planes: microsoft-agent-governance-toolkit - counters refreshed (165 open issues, push 2026-11-17); no release past v4.1.0.
- control-planes: mission-control - default-branch push corrected to 2026-10-28 (the 2026-11-09 and 2026-11-17 repo pushes side branches); counters refreshed.
- control-planes: okto-pulse - the v0.4.3 tag landed 2026-11-17 while PyPI still publishes 0.4.2.
- control-planes: openappa - v0.43.0's release page published; v0.43.1 (2026-11-16) a dependency-only patch.
- control-planes: paperclip - backlog to 6,863 pushed 2026-11-18; the release pause reaching day twelve of v2026.1106.0.
- control-planes: settlebridge - 278 to 314 commits on main: per-purpose mediator-call secrets, the verdict rule moved into the private verifier, operational runbooks moved out of the repository.
- control-planes: speakeasy-gram - release train to server@2.36.0 (2026-11-18): hosted MCP servers endpoint-only with per-server-scoped tokens, organization encryption required for OAuth-credential processes, upstream auth enforcement default-on for new organizations.
- control-planes: tracekit - rotation re-verification clean; release train to v1.7.2 (2026-11-18, PyPI in step).
- control-planes: vestige - open issues and pull requests 235 to 9 on the maintainer's bulk close of its own wave-tracking issues (2026-11-18); no new release past v6.2.2.
- control-planes: matrix volatile cells refreshed, no membership change; the category index unchanged with membership stable at twenty-four.
- control-planes: the awesome-ai-governance source unchanged since 2026-11-16, its scan carried over with no new entrants.
- evaluation-review: mapped source anhermon/awesome-agent-observability rescanned through its 2026-11-16 commits (the 2026-11-15 evals expansion, DECLINED.md, PyRIT and PurpleLlama, the mcp-trace removal) with three admissions
- evaluation-review: swe-bench - Created from the evals-expansion scan as the ground resolved-issue benchmark, the SWE-Bench+ leakage critique and the 466-point Git-history thread recorded
- evaluation-review: harbor - Created as the neutral benchmark runner and Terminal-Bench 2.0 official harness, the Runta provider overlap with FrontierHarness recorded
- evaluation-review: tau2-bench - Created as the dual-control service-agent benchmark, the 197-point prompt-sensitivity thread as the critical source
- evaluation-review: langfuse - Recorded the v4.81.0 (2026-11-16) and v4.82.0 (2026-11-17) releases and refreshed the activity line
- evaluation-review: phoenix - Recorded the arize-phoenix 21.2.0 release (2026-11-18) and refreshed the activity line
- evaluation-review: plannotator - Recorded the v0.28.66, v0.28.67, and v0.29.0 releases (2026-11-16 to 11-18) and refreshed the activity line
- evaluation-review: pr-lens - Recorded the npm CLI at 0.15.2 (2026-11-16, renderer 0.6.3 looping walkthrough embed) and the Unicode-filename fix
- evaluation-review: reviewbench - Recorded the Open Code Review registration removal (PR 115) and the 83-finding golden-set cull (PR 107), both 2026-11-16, twenty-three registrations standing
- evaluation-review: verbatim-citation-gate - Recorded the unreleased 2026-11-16 gate fix (operator signs count), v0.3.2 still newest
- evaluation-review: better-harness - Rotation re-verified, every cited URL live, the Status pushed date corrected to 2026-11-11 (the PR 215 merge)
- evaluation-review: harnesstax - Rotation re-verified, every cited URL live, two more AgentBRANE dashboard-build republications recorded (2026-11-13)
- evaluation-review: hermes-gate - Rotation re-verified, every cited URL live, the 2026-11-13 push identified as another dependabot bump (PR 61)
- evaluation-review: crit - Activity line refreshed (pushed 2026-11-17, CSP remote-image fix, CLI branch inference), no release change
- evaluation-review: deepeval - Activity line refreshed (pushed 2026-11-18, voice duplex fixes and revert), 4.3.4 still newest
- evaluation-review: hunk - Activity line refreshed (pushed 2026-11-17, unreleased web review-UI stream), v0.25.0 still newest
- evaluation-review: kiln - Activity line refreshed (pushed 2026-11-18, model-catalog additions), v1.2.1 still newest
- evaluation-review: kitaru - The 2026-11-17 repository-level push identified as side-branch, default branch unmoved
- evaluation-review: frontierharness-eval - The 2026-11-17 repository-level push identified as side-branch, default branch unmoved
- evaluation-review: matrix extended to twenty-five columns (Harbor, SWE-bench, τ²-bench) and maturity cells refreshed (Langfuse v4.82.0, Phoenix v21.2.0, Plannotator v0.29.0, PR Lens age)
- evaluation-review: category index updated to twenty-five entries with three additions
