## (daily refresh)

- Added 24 research notes with sorted matrix columns, category-index entries, and Added bullets: Copilot CLI, Command Code, and Omnigent (harnesses); SkillMD (skills); open-codebase-index and Orama (retrieval); ccusage and CodeBurn (session analytics); Agent Control and Jamf AI Governance (control planes); OpenResearch (automated research); ANP (protocols); CodeAlive (context engines); Jeeves and Ollaya (hybrid execution); AI-DLC and cc-sdd (spec-driven development); Agent Manager, LangGraph, Raven, Ruflo, and Scion (orchestration); Jevals (evaluation and review); Magnitude (model access) [glm-5.3-flash]
- Rejected the AMO directory's uncovered entries (ccswarm, solveathome/platform, coven, orbi, workkit, claudexor, 5dive, artifact-council-relay, OrcaReplay, and the standing tail) with reasons [glm-5.3-flash]
- Rejected the category workers' candidates with reasons, including the skills directory wave (skillsdirectory.com, agentskill.sh, agensi.io and peers), the memory and session-analytics wave, the control-planes and software-factory closed products, and the people first-surfaced test [glm-5.3-flash]
- kimi-k3: recorded its fifth move in five days to $0.95/$14.00 (cache $0.31) with the Relace floor at $0.96/$13.00 and a Price history row; cursor: corrected Teams Premium to $120/user/month and added the India-only Start plan [glm-5.3-flash]
- Recorded openchamber v2.1.1, openrig at 5.0k stars with v0.6.5, orca v1.4.220 at 85.2k, squad v1.0.1, and agent-swarm v1.163.0, with seventeen matrix status cells moved [glm-5.3-flash]
- Recorded AG-UI at about 15.0M monthly downloads, SkillMD's 1,948-of-1.13M safety-review gap, Cursor's new native Agent Skills documentation, claude-mem v13.31.0, agenttrace v0.10.0, and memex v0.27.1 [glm-5.3-flash]
- Recorded open-code-review v1.12.12, langfuse v4.51.0, paperclip at 97.3k, OpenShell's surge at 14,917, and clawk's quiet window at fifty-three days; every stated model-access price re-fetched unchanged [glm-5.3-flash]
- Recorded Anthropic's September-October science run, corrected OpenAI for Science's stale Clay claim, and noted Karpathy's Bear blog back up and four YouTube RSS feeds serving again [glm-5.3-flash]
- model-selection: recorded the Kimi K3 listing and host-floor moves and corrected Sonnet 5's release date per models.dev; the landscape re-verified its star chain (Orca about 85.9k, OpenHands about 90.1k, DeepSeek Harness about 244.1k, Claude Code about 149.5k, OpenCode about 211.9k) [glm-5.3-flash]
- Brought the fourteen membership changes current in agents/_index.md; no new queue questions and no self-directed essay [glm-5.3-flash]