## (daily refresh)

- Added 52 research notes across the categories (automated research 17 to 26, orchestration 66 to 69, memory 23 to 27, surfaces 17 to 23, hybrid execution 20 to 26, and others) with matrix columns and index entries [glm-5.3-flash]
- kimi-code: recorded the kimi-k3 move to $0.80/$10.00 (the eleventh move), and gpt-5.6-terra and luna returned to OpenAI's pricing page [glm-5.3-flash]
- AutoGPT restructured hosted pricing, cmux brought yearly billing back, openwork's pricing churned a thirteenth time, and experiential's discount schedule expanded [glm-5.3-flash]
- open-swarm: recorded the pivot to a proprietary Open Swarm Inc.; github-agentic-workflows gained a sixth built-in engine [glm-5.3-flash]
- Added the star-history charts to 16 of the day's new single-repository notes, and re-sorted the skills matrix columns around a misplaced SkillSpector plus the protocols category index [glm-5.3-flash]
- Rejected deepagents (an agent-building SDK), rea.tools (skills-kind, held under the pack exclusion), and the promptfoo, Opik, and AgentOps holds; no new voice passed the first-surfaced test [glm-5.3-flash]
- Section roots: updated model-selection with the Kimi K3 floor's same-day movement and terra/luna's return, and re-dated the landscape, ADE tracker, and benchmark matrix [glm-5.3-flash]
- Queue: appended six new `? for tom:` questions [glm-5.3-flash]