WorkSwarm adds personal knowledge spaces: agent task pass rate up 32%, token usage down 84%
量子位 · wechat · 2026-10-13
WorkSwarm (openJiuwen) launched a "personal knowledge space" that turns scattered personal materials into reusable background knowledge for office agents, eliminating the need to re-gather context for every task.
Workflow: Users configure sources like local folders, Edge bookmarks, Zhihu columns, Feishu docs and GitHub; the system collects incrementally on demand or on schedule, then organizes content via three modes—rule-based (BM25), hybrid rule+LLM, or agent-driven—into categorized Markdown pages with cross-links and citations. At task time, progressive disclosure lets the agent read an overview first, then follow links as needed.
Results: On a 25-question internal benchmark across seven source types, task pass rate rose 32%, total task time fell 76.2%, and task-phase token consumption dropped 84.1%. Available on the official site, GitHub and AtomGit.
More from coding & agent
- Laude Institute launches FrontierCS 2: agents must beat deleted paper contributions blind — lateinteraction · 2026-10-23
- Index Network open-sources Social Discovery Protocol: agents negotiate fit on your behalf — carsonfarmer · 2026-10-23
- Installed a formal-verification skill: it silently installed Java, MacBook runs 10 degrees hotter — zealcaiden · 2026-10-23
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- Ruff author warns: upgrade Node/Bun now — --use-system-ca still accepts the eMudhra cert — charliermarsh · 2026-10-23
- Proof of Agent: paste one prompt and your AI mines, grades, and self-governs — adamamcbride · 2026-10-23