@jinchenma_ai open-sources 11 skills for an agent-maintained personal knowledge base: originals protected, sources traceable, Feishu recordings optionally synced
The biggest trap building a knowledge base is copying someone else's heavy stack. Chinese author @jinchenma_ai open-sourced jinchenma-knowledge-base: rules in AGENTS.md, 11 skills (capture, media-ingest, meeting-ingestion, ai-dialogue-ingest, knowledge-analysis, query, enrich, entity-propagation, …) for notes, bookmarks, meetings, full AI chats, candidate insights (human confirm), query and entity links; optional Feishu recording sync and ASR. License CC BY-NC 4.0 (non-commercial). On the box with pytest/PyYAML, `scripts/run_tests.py`: 146 passed + 8 subtests; `check_project.py` reported 11 skills and no broken links; `setup.py --mode create` built a full vault template in a temp dir. No real Feishu account or end-to-end ingest.
Visit site jinchenma-knowledge-base (Jinchenma Knowledge Base) ↗- Pricing
- Free
- Region
- Global
- Where it's from
- Chinese community
- Added
- Nov 6, 2026
- Tested on
- Nov 6, 2026
- Version we used
- jinchenma-knowledge-base @ d9c0bda (2026-11-04); 11 skills per manifest; pytest via scripts/run_tests.py
- Categories
- ProductivityResearchWriting
- Original author
- 金尘马 @jinchenma_ai ↗
- Our test
- Partly tested
- Cost / time
- Repo is free/open, but CC BY-NC 4.0 bars commercial use. You still need a local agent subscription/quota; optional Feishu, ASR, WeChat/Douyin ingest need their CLIs, cookies or paid APIs.
The task
Use a local agent (WorkBuddy / Codex / Claude Code) to turn diaries, bookmarks, meetings and AI chats into a traceable, queryable personal knowledge base—not a pile of loose Markdown.
Steps
- Clone https://github.com/jinchenma94/jinchenma-knowledge-base (Python 3.10+). Treat the repo as the agent working directory and read AGENTS.md; skills run from this project—no global skills install required.
- Tell the agent: read AGENTS.md and create the vault in a separate folder; confirm path, timezone and diary day boundary first. Or run: `python3 scripts/setup.py --vault-root /abs/path/to/vault --mode create --timezone Asia/Shanghai --day-start-hour 6`. Use `--mode attach` for an existing vault (no migration); vault must not sit inside this repo.
- Daily triggers: capture for notes; media-ingest for bookmarks; meeting-ingestion for transcripts; ai-dialogue-ingest for full AI chats; knowledge-analysis for candidate insights (confirm before write); query to look things up.
- Optional Feishu: `pip install -r requirements-feishu.txt`, install/login lark-cli, `setup.py --check feishu`, then `--feishu-start-from`. Keep credentials local; nothing is written back to Feishu.
- License: CC BY-NC 4.0, non-commercial only; attribute and note changes if you republish. Don't copy the folder layout blindly—trim to your materials.
Prompt
Please read this project's AGENTS.md and set up my knowledge base in a separate folder I name—confirm the path, timezone and diary day boundary first. Project: https://github.com/jinchenma94/jinchenma-knowledge-base
What the author says
@jinchenma_ai: don't start by copying a heavy architecture; three content kinds are originals, published outputs, and compiled knowledge (conceptual—not three mandatory folders). The open kit ships a vault template and 11 core skills; never mutate originals; the compile layer may iterate with provenance; personal opinions need human confirmation. Works with WorkBuddy, Codex or Claude Code.
Our test
Cloned jinchenma94/jinchenma-knowledge-base (commit d9c0bda). After pytest + requirements-dev: scripts/run_tests.py and scripts/check_project.py; setup.py --mode create on a temp vault and --check base. No Feishu login, no live ASR/media ingest, no full dialogue ingest inside WorkBuddy/Claude Code.
What we found when we tried it:pytest: 146 passed, 8 subtests passed (~2.1s). check_project.py: ok, skill_count=11, errors=[]. setup create: status=configured; temp vault had 00-inbox…99-config dirs plus AGENTS.md; --check base → ready.
How to get in
Public GitHub repo; needs git/Python. Tencent WorkBuddy works for CN users reading a local vault; Claude Code / Codex also fine. Original post and docs are mostly Chinese.
Who it suits
People who want an agent to steward personal materials, writing sources and meeting notes long-term; OK with a non-commercial license and confirming insights before they land.
Pitfalls
- License is CC BY-NC 4.0—commercial bundling may be out of bounds; read LICENSE and NOTICE.
- Vault must live outside the project; `create` only accepts an empty dir—don't commit personal notes into the public kit repo.
- knowledge-analysis proposes candidate insights—confirm before write; agents may misread the three content kinds as three folders.
- Our run: 146+8 unit tests passed and vault template created; no live Feishu account—remote sync quality still depends on your machine.
More in this category
- Keep a long-running AI-built project from turning into a messCodingClaudeCodexoil-delegateNeeds Claude and Codex plans◐ Partly testedChina
- Have fresh-context AI reviewers poke holes before you codeCodingEGMCodexClaude CodePlugin free; reviews use your model quota◐ Partly testedOverseas
- 1,050 memory recalls audited: what makes agent memory pay offCodingNowledge MemCodexClaude CodeFree tier: 50 memories; review uses model quota○ Not tested by usChina
- Isolate an always-on agent on its own machine, then red-team itAutomation & AgentsOpenClawClaude CodeOllama$20/mo Ollama Cloud + a Claude plan○ Not tested by usOverseas
- A to-do list beside your agents, one AI chat per taskProductivityToday BoardcmuxClaude CodeFree, open source; needs macOS + cmux◐ Partly testedChina