
## (daily refresh)

- Recorded the day's status refreshes across harnesses, surfaces, and orchestration (claude-code open issues and PRs down to about 13.2k, cline installs 5,234,440, deepseek-harness 214.4k, roo-code's archived drift at 24,309 stars); reverted three notes' over-set `updated` dates [glm-5.3-flash]
- openwork: recorded Enterprise at $40/user/month and the Team Starter rename to Team; nanobot issues at 770; instructor about 24M monthly PyPI downloads [glm-5.3-flash]
- skills-sh: recorded the leaderboard move (open.feishu.cn 16.8M, runcomfy 11.0M, larksuite about 10.9M, azure-validate back marked Safe); github-agentic-workflows v0.88.5; ouroboros v0.54.0 [glm-5.3-flash]
- landscape: folded the Orca count in (about 60k to about 63k stars) and bumped to 2026-09-07; model-selection re-fetched all eleven references with zero price moves [glm-5.3-flash]
- Matrices: all 20 re-verified and re-dated to 2026-09-07 with cells updated where facts moved (orchestration Omnara stars, assistant-runtimes Hermes and PicoClaw, retrieval LangChain stars, evaluation FrontierHarness 82 points, and others) [glm-5.3-flash]
- Rejected with evidence: engrim (far below the memory category's entry bar), Ponytail (a community skill out of Skills scope), and MathKernel (a domain-specific mathematics MCP server); GitHub sweeps surfaced nothing credible [glm-5.3-flash]
- agents/_index.md: re-dated eleven one-liners to 2026-09-07; the three `? for tom:` questions remain unanswered and no self-directed essay ran [glm-5.3-flash]
