## (daily refresh, 1)

- assistant-runtimes: langbot - created the note from the elyase/awesome-personal-ai-assistants scan: Apache-2.0 IM-bot platform, 18,601 stars as of 2026-10-30, v4.11.4 (2026-10-29), PyPI Production/Stable, zero English-language HN footprint recorded as the signal.
- assistant-runtimes: assistant-runtimes - feature matrix gained the LangBot column (sixteen members, re-sorted alphabetically, oldest-member prose corrected to LangBot/AstrBot), with drifted cells refreshed (AstrBot 1,109, Octop 900, Open WebUI v0.12.5, OpenClaw v2026.10.9 with 7.9k, PicoClaw 37, PrivateGPT pushed 2026-10-29) and the index entry added.
- assistant-runtimes: open-webui - recorded v0.12.5 (2026-10-29: admin analytics session explorer, multi-page notes with live presence, another unlisted security-fix set pending CVE assignment).
- assistant-runtimes: openclaw - recorded v2026.10.9 (2026-10-29, security-dependency and recovery train) taking the latest-stable spot beside the v2026.9.36 extended-stable patch, with the v2026.10.10 line at beta.5.
- assistant-runtimes: openwork - pricing churned an eighteenth time (the two add-ons returned, the grandfather clause left again; Price history row).
- assistant-runtimes: astrbot - rotation re-verification: v4.29.3 still latest, every cited URL live, the QQ-adapter issue 8131 and the zero-HN-footprint record intact.
- assistant-runtimes: picoclaw - rotation re-verification: v0.3.1 (2026-07-03) still latest, both README banners intact, picoclaw.io and docs serving with valid TLS.
- assistant-runtimes: assistant-runtimes - volatile refresh (issue counts and push dates) across AnythingLLM, Eigent, Hermes, Nanobot, NanoClaw, Octop, OpenWorker, PrivateGPT, QwenPaw, and ZeroClaw.
- agon: rotation re-verification, all seven cited URLs live, facts unchanged (repo last push 2026-10-06, HN footprint still zero hits), no note change
- autocontext: rotation re-verification, all eleven cited URLs live, PyPI still 0.18.0 and npm still 0.19.1 with the README pin unchanged, no note change
- data-to-paper: rotation re-verification, all seven cited URLs live, still dormant at the 2025-07-19 push, no note change
- coral: rotation re-verification, all seven cited URLs live, main and the release train still at September 6 (v0.7.21), no note change
- local-deep-research: rotation re-verification, all seven cited URLs live, status push date moved to 2026-10-30, open issues and PRs 1,267 to 1,282, release train and PyPI still v1.10.7
- aris: volatile re-verification, the October 29 docs-only push (search auth and DeepXiv env docs) was already recorded, v0.4.28 still the latest release, no note change
- autosci: volatile re-verification, October 30 push is another WeChat QR docs refresh, status push date moved to 2026-10-30
- kapso: volatile re-verification, October 29 commit batch recorded (kapso report CLI export, windowed repository-memory update prompts, three fixes), release train still v0.7.10
- openresearch: volatile re-verification, October 29-30 post-release commits recorded (native skill discovery, Windows and Linux deep links, Base UI dialog system), v0.2.22 still latest
- automated-research: link, front-matter, matrix, and category-index checks clean across all 34 notes; membership unchanged, no matrix or index edit
- automated-research: dedicated-source check found yenanjing/awesome-ai-for-science (12 stars, 502 entries, updated 2026-10-29), wiring decision sits with q#26
- multi: rotation re-verification recorded v1.20.0 and v1.20.1 (October 22), ponytail-5 defect triage, load-scale review scope, and per-reviewer key isolation, tag count 28; eval page and zero-HN footprint unchanged
- git-lrc: rotation re-verification re-fetched every cited URL, prices and release state unchanged (v0.6.17 latest, 55 releases), and recorded the Free tier as the pricing page now reads (hosted LiveReview AI included beside BYOK, 3-member cap, risk-based auto-approve) plus the pricing FAQ's no-SOC-2 statement
- kodus: volatile re-verification recorded the October 29 release batch (2.2.9, web-1.2.10, selfhosted-2.2.7, mcp-manager-2.1.5) with conditional PR approval keyed on historical code-area risk and an EE-license clarification of Community self-hosted production rights; push moved to October 30, prices unchanged
- open-code-review: push moved to October 30 with no release after v1.13.3
- openqodex: recorded openqodex@0.16.0 (October 28), JUnit test-method linking and the possible-issues split, npm at 8,685 trailing-month downloads to October 30
- qodo: push moved to October 29 with v0.49.0 still the latest release
- roborev: push moved to October 30 with v0.74.0 still the latest release
- code-review: refreshed the matrix's OpenQodex npm cell to 8,685 trailing-month downloads; other flagged members' cells unchanged
- code-review: kodustech/awesome-ai-code-review re-check confirms the vendor list still frozen at its June 28, 2026 push (q21 evidence refreshed); no credible dedicated code-review list has appeared (largest dedicated candidates 1 to 2 stars)
- context-engines: semble - Recorded release v0.7.1 (the skill option shipping, an opt-in HTTP transport on the MCP server, a clearer Hub-download error) and refreshed forks and push date; no-paid-plan claim unchanged.
- context-engines: wrenai - Recorded wren-v0.17.1 (the DuckDB 2.0 to_arrow_table fix) with PyPI tracking 0.17.1; volatile numbers refreshed and credit tiers re-verified unchanged.
- context-engines: graft - Recorded npm 0.27.0 and 0.28.0 (one-time-code agent sign-in, learnings before an edit, the pre-commit trail check warning, multi-workspace laptops) and refreshed growth numbers; Trail Brain pricing re-verified unchanged.
- context-engines: serena - Rotation re-verification: every cited URL intact, JetBrains installs refreshed to about 20.1k, forks and push date refreshed, stale PyPI reference line corrected; no claim changed.
- context-engines: code-review-graph - Rotation re-verification: every cited URL intact, forks and open issues refreshed; no claim changed.
- context-engines: vanna - Rotation re-verification: every cited URL intact, three pricing tiers unchanged, forks refreshed; no claim changed.
- context-engines: mapped source Meirtz/Awesome-Context-Engineering unchanged since 2026-10-27, scan carried over, no additions to resolve.
- context-engines: pushed-signal checks resolved six members as non-material (code-review-graph and SoL-Pi branch-only pushes, jevgrep skill rename, repomix fixes, repowise fixes and internal seams, headroom fixes and docs, rtk fixes and dev RCs, entroly post-release fix), notes untouched per the 2026-10-25 ruling and claim checks.
- vestige: v6.0.0 (2026-10-28) ships Reflex, carrying every hook firing into the server as stop memories over a hash-chained Hook Ledger, and Operator Lite absorbs every per-call rule the paid gate had (42 rules, 27 STOP, planted canaries included)
- vestige: paid Operator now sells laws, the Board, the weekly Letter, onboarding, and the cross-session stop record; price unchanged at $149 once
- tracekit: v1.5.0 (2026-10-28) adds experimental cross-agent surfaces (A2A gateway with C2C transfer records, PCM 0.1 proof-carrying messages, tracekit graph) and goose, Qwen Code, Antigravity, and Factory Droid hooks
- tracekit: README proofs moved to 699 of 699 harmful calls stopped, and the project's arXiv paper (2609.35659) added, itself conceding 211 of 420 on the blind pool
- arifos: self-measurement re-stamped 2026-10-29 at 12 of 13 floors measured pass with F6 unmeasured and none failing, kernel still degraded, independent audit unchanged
- arifos: model-governance card landed on init envelopes (server-minted model provenance, fail-closed UNKNOWN, enforcement surface 32 to 33)
- okto-pulse: PyPI reached 0.4.2 (2026-10-28, patched Authlib and filelock pins) and okto-nexus reached 0.3.2
- okto-pulse: product page caught up to v0.4.2 with 280+ MCP tools, resolving the v0.2.6 lag and the 215-versus-285 tool-count disagreement
- agentfield: the built-in coding harness renamed CodeAF (AForge alias) with Pi and Oh My Pi added as providers; candidate line reopened at v0.1.146-rc.3
- crowdstrike-falcon-guardian: rotation re-verification found no drift
- openappa: rotation re-verification found no drift
- control-planes: matrix volatile cells refreshed, no membership change

## (daily refresh, 2)

- langfuse: recorded the v4.72.0 release (2026-10-29) and refreshed the activity line
- phoenix: recorded the arize-phoenix 20.27.0 (2026-10-29) and 21.0.0 (2026-10-30) releases, the first major bump, shipping a MIGRATION.md and turning auth on by default, and refreshed the activity line
- plannotator: recorded the v0.28.36 (2026-10-28) and v0.28.37 and v0.28.38 (both 2026-10-29) releases and refreshed the activity line
- slop-grader: recorded the v0.5.0 release (2026-10-29, standard-input reading with `-`) and refreshed the activity line
- pr-lens: recorded the npm CLI at 0.14.0 (2026-10-27), whose train added the data-model lens, plus the merged CLI security hardening, and refreshed the activity line
- jevals: rotation re-verified with every cited URL live; JevBench board refreshed from 196 to 235 ranked open-weights systems, Jev 1.14.0 now a second unranked reference (capability 75.0)
- better-harness: rotation re-verified with every cited URL live; near-daily-push claim corrected to a stop on 2026-10-20, npm stable line unchanged at 0.6.6
- frontierharness-eval: rotation re-verified with every cited URL re-fetched and live; repository unchanged (last push 2026-09-08)
- evaluation-review: volatile re-verification untriggered for agent-qa, deepeval, hermes-gate, hunk, kiln, kitaru, and reviewbench (pushes without releases or material default-branch changes; deepeval PyPI still 4.3.2, reviewbench onboardings still twenty)
- evaluation-review: feature matrix maturity cells refreshed; membership, sort order, and category index list unchanged at 21
- trigger-dev: recorded the v4.8.2 release (2026-10-29) and the public-beta inbound-webhooks launch (2026-10-26, 40+ named providers including GitHub, Slack, and Stripe, Standard Webhooks and Svix support, signature verification, dedup, up to 25 agents per event); rotation re-verify re-fetched every cited URL, pricing tiers and the self-hosted cloud-only feature table unchanged; forks 1,522, pushed 2026-10-29.
- executions-feature-matrix: Trigger.dev trigger cell and trigger prose updated for the webhooks beta (2026-10-26); no membership change.
- codex-action: rotation re-verify with every cited URL re-fetched; v1.13 still the newest tag, changelog commits through 2026-10-20, permission profiles still beta at CLI 0.138.0+, permissions reference live at learn.chatgpt.com, still no dedicated HN thread; counters unchanged (181 forks, 82 open).
- claude-code-action: release train refreshed to v1.0.265 (2026-10-30); counters 2,233 forks, 850 open issues and pull requests; the Latest badge still sits on the v1.0 announcement release.
- github-agentic-workflows: release train refreshed to v1.0.1 (2026-10-27, stable-marked) above the v1.0.0 prerelease; README advisory notice still live, no new advisory since the October 14 batch; forks 609.
- claude-code-hooks: docs re-verified after the v2.1.314 release; the 33-event table, five handler types, and the v2.1.267 and v2.1.295 version gates all still present on the live page; open issues roughly 16.1k.
- aeon: counters refreshed (293 forks, 1,509 commits, pushed 2026-10-29); proof-of-work page moved to 148 findings across 110 repositories; the single v0.1.0 release re-confirmed.
- airut: pushed 2026-10-30 with no release since v0.27.1; 287 trailing-month PyPI downloads re-dated to the current window.
- rivet-agents: counters refreshed (136 forks, pushed 2026-10-29 UTC); v0.5.4 still the newest release.
- n8n: volatile check unchanged (forks still 61.3k at displayed precision, no tracked claim moved on the n8n@2.45.6 release).
- executions: no credible dedicated source appeared (GitHub search returned zero-star stubs only).
- pi: rotation re-verification, every cited reference live; recorded Pi Durable, the experimental durable-harness framework shipped beside Pi 1.0, and the port ecosystem (pi_agent_rust at about 1.9k stars, pz dormant), with v1.4.0 still the newest release.
- grok-build: rotation re-verification, every cited reference live except the changelog (still 403s); alpha dist-tag moved to v1.0.73 (October 28) with stable holding at v1.0.70; monthly downloads 260,703 (September 30 to October 29); API prices unchanged.
- jcode: rotation re-verification, every cited reference live; the pricing page reworked the automatic 10x-the-plan usage cap into a default-$0 extra-usage limit and moved justrust builds to 2,500 per $10 of plan from the $20 tier (Price history row); the live prompt counter reads 248 tokens.
- longhorizon-harness: rotation re-verification, every cited reference live, v0.1.8 still newest with no commits since October 21; pypistats rolling month re-read at 1,048.
- zerostack: rotation re-verification, every cited reference live; crates.io counters to 3,760 total (724 recent), v1.8.5 still newest.
- amp: rotation re-verification, every cited reference live except the Use Your Claude Plan announcement, whose URL now serves the pricing page; recorded the October 27-29 news (Android app, personal self-hosted orb providers, thread visualization, macOS Satellites beta) with the satellites' Price history row; tier page unchanged.
- command-code: rotation re-verification, every cited reference live; npm train reached 1.88.2 (October 29), changelog counts 468 releases; GOAT credits raised from $70 to up to $100 and Pro from $80 to up to $110 (Price history rows); downloads 157,393 (September 30 to October 29).
- crush: recorded v0.99.0 (October 29), the Catwalk v3 provider-model rework with run-wide hooks and a client/server mode; commits refreshed to 4,390.
- fx: recorded v0.0.16 (October 29), the token-and-size release with the 60-percent-lighter libfx install and 16 MiB file edits.
- qwen-code: recorded v0.26.4 (October 29) with sdk-typescript-v0.1.26 and desktop-v0.26.4 the same day; push date to 2026-10-30.
- deepseek-harness: recorded dsh-v0.2.1-rc.5 (October 30) taking the newest-tag spot above the 0.2.2 alpha train; still prerelease-only.
- oh-my-pi: npm latest reached 18.14.0 (October 29); push date to 2026-10-30.
- claude-code: recorded v2.1.314 (October 30), blockWritesOutsideWorkingDirectories and idle compaction headline the release.
- harnesses: push-date refreshes recorded for avante-nvim, exo, gemini-cli (commits to 6,562), gptme, kilo-code, mimo-code, nanocoder (npm to 7,950 monthly), om-o, omnigent, openhands, and zcode.
- harnesses: feature matrix and index membership re-checked (45 members, sorted, all rows at header width); DeepSeek Harness index one-liner moved to the rc.5 pin.
- jeeves: rotation re-verification corrected the thinking-latency trade to the v2-era figures (20 ms direct, 1.2 s median with the confidence gate, 1.9 s when every question reasons, on one H100 fp8), the stale v1-era 3.3 s retired; recorded the README citation block, the jeeves-fp8 weights repository, and the v1-levels-v2 limitation; weights 356 downloads as of 2026-10-30.
- marvin: release train resumed 2026-10-27 with three same-day releases (v3.2.8 through v3.2.10, dependency and docs catch-up, PyPI 3.2.10), ending the March-to-October drought; status and caution rewritten around catch-up-not-momentum; pushed 2026-10-27.
- laya: 0.4.16 published and v0.4.17 released 2026-10-28 (Go router lifecycle with SHA-256 pins, C++ port foundation, .NET calibration ports, Vietnamese routing), PyPI 0.4.17, 0.4.18 release commit on main unpublished; pushed 2026-10-30.
- valen: first tagged release v0.3.0 (2026-10-28): CHANGELOG, release process, SDK publications, pinned gguf-runtime packaging; valen-sdk 0.3.0 in the README, Valen.llm() and agent-skill announcements, Jev-ToM community evaluation; pushed 2026-10-29.
- baml: 0.23.1 nightly train through the 20261028.a build (published 2026-10-29), open issues 360, pushed 2026-10-30.
- atomic-agents: pushed 2026-10-29 (MCP scalar enum/const fix #300, agent-mode derivation #293, quickstart model-ID replacement #331), no release since v2.15.0.
- guardrails: pushed 2026-10-29 on branch activity, v0.11.0 still latest.
- ollaya: jev-style and decision2 families merged 2026-10-26, not yet on the site (count re-read 29 on the 2026-10-30 fetch); pushed 2026-10-30.
- strands-decider-2b: pushed 2026-10-29 on branch activity, v0.2.0 still latest.
- tool-eval-bench: pushed 2026-10-28 with the unreleased toolscale sweep (#324, #325) and run-deletion work (#326).
- cua-s1: host pushed 2026-10-30; the cua-spaces-v0.18.0 and v0.19.0 releases do not touch the cua-s1 component.
- anthropic-structured-outputs: rotation re-verification, all five cited URLs live, supported-model list and token table unchanged.
- instructor: rotation re-verification, all five cited URLs live, PyPI 1.17.0 still latest, monthly downloads flat at 8,516,567 for a fourth check.
- outlines: rotation re-verification, all seven cited URLs live, repository unchanged since 2026-09-21.
- hybrid-execution: matrix maintenance cells moved for the eleven signal members, Jeeves latency cells re-based on the v2-era figures; matrix membership, sort, widths, category index, links, front matter, and chart markup all verified clean.
- basic-memory: recorded the v1.0.0 release (October 29, PyPI in sync): the bm_* MCP tool line shipped with shared parameter types and typed refusals, plus a fix refusing note writes whose path leaves the project.
- claude-mem: recorded v15.6.0 (October 29): My Work state unified in one Zustand store with URL-addressable views and remembered filters; npm in sync at 15.6.0, trailing-month downloads 75,512.
- cognee: recorded v1.6.6 stable (October 29): durable fact conflict-resolution API and default-on conflict review in the open engine, re-embedding-free vector copying, atomic Redis deletions.
- cognee: matrix contradiction cell moved to the open-engine conflict resolution (bi-temporal memory stays Enterprise BYOC only).
- deja: recorded v0.23.3 (October 29): test-file classification in FinalChanges, library edits sorting ahead of `*_test.py` edits in recalled code changes.
- memanto: recorded v0.2.31 (October 29): caller attribution unified around --source (breaking), connect instruction and skill path fixes, session-cache and OKF import fixes; npm back in sync.
- gbrain: recorded the release train's continuation to v0.60.505.0 (twenty-six releases, October 28-30): the visibility-private spelling hole closed (previously world-readable), provenance no longer remote-rewritable, the brain-first rule riding every turn in every harness install, search superseding corrected facts (27/27 replay), dated-saved-facts lead in budgeted queries (183/200 versus 151).
- hindsight: extended the pending-release work (October 28-29): response-schema tool calls with native tool calling dropped, small-model consolidation, a hosted decision-model provider with sign-in, recall rerank prune and dedup flags, coding-agents v0.8.3.
- funes: recorded the pending-release work since v1.9.1 (October 27 and 29): per-memory language counts in status, push, and miss lines, and reranking extended to any pool larger than the hits.
- memory: memU: recorded the further pending retrieval commit (October 29): backend-specific score cutoffs.
- memos: recorded the pending-release plugin commits (October 29): SQLite 13 prebuilds with Node 22 required (breaking), Windows DLL isolation, DSH memory-issue fixes.
- openviking: recorded the pending-release main train (October 29-30): native BM25 for openGauss, breaking batched group-member administration, an opt-in Redis directory TTL, Studio terminal filesystem commands, compression preserving retained previews, a Hermes recovery-lock handoff fix, agent onboarding guides.
- screenpipe: recorded the app train to v2.7.132 (October 28-30): upload-starvation repair during retention, interrupted Windows migration retries recovered, excluded windows kept out of screenshots, no-training Experiential routing carrying the user's training-consent choice.
- supermemory: recorded the pending-release fixes (October 28-29): memory-core enforcing agent and task read permissions, dreaming status exposed over MCP, finite transport timeouts required in the Python SDK.
- agentmemory: rotation re-verification, every cited URL re-fetched live, no facts changed; post-release pushes are a Cline MCP settings-path fix, doctor diagnostics, and viewer docs.
- emulo: rotation re-verification, every cited URL re-fetched live, no facts changed (repository quiet since 2026-10-12, PyPI at 0.6.7, both study pages unchanged).
- engrim: rotation re-verification, every cited URL re-fetched live, no facts changed (no release after v1.4.11, the Show HN thread steady at 93 points).
- letta: rotation re-verification, every cited URL re-fetched live, pricing ladder unchanged (Free 3 stateful agents, Pro $20/mo); letta-code's latest tagged release refreshed to v0.34.33 (2026-10-30).
- operator-memory: rotation re-verification, every cited URL re-fetched live, no facts changed (helper still 1.9.0 with twenty releases, the essay thread steady at 392 points).
- memory: mapped source unchanged since its 2026-10-29 check (last commit 2026-10-27), scan carried over with zero new candidates.