ICCuse/dsh-knowledge

Bridge into a global Markdown knowledge base shared with the Codex kb.cmd CLI: kb_add/kb_search/kb_show/kb_timeline tools with byte-compatible frontmatter.

Bridges the agent into the user's global knowledge base — the same Markdown + YAML frontmatter library the Codex kb.cmd CLI writes, so knowledge accumulated by either tool is visible to both: one library, two agents. Four tools: kb_add (write one typed note — lesson/pattern/preference/project/reference/concept/moc/decision/note — byte-compatible with the Python CLI: slugified unique filename, typed directory, YAML frontmatter), kb_search (keyword scan over notes, title > tags > content), kb_show (locate a note by title/path/tag and return full text), kb_timeline (append a timestamped line to notes/99_时间线.md after every writeback). The heavy read side (ask/reflect/map/ingest) stays on the Python CLI.

Memory ★ 2 updated 2026-08-14
View on GitHub ↗

Install

npm install github:ICCuse/dsh-knowledge

GitHub install per README (English): npm install github:ICCuse/dsh-knowledge (or pnpm add github:ICCuse/dsh-knowledge), then mount the bundle (declared in package.json 'dsh.bundle') and point it at your knowledge base with a cordis.patch.yml row: - id: knowledge / name: 'dsh-knowledge' / config: kbRoot: 'D:\knowledge'. Not on npm per README as of 2026-08-28.

Compatibility

DeepSeek Harness. Bridges to the same Markdown + YAML frontmatter knowledge base the Codex kb.cmd CLI writes (default D:\knowledge).

Details

Recent updates

global knowledge base bridge; kb_add/search/show/timeline tools; byte-compatible with kb.cmd CLI; one library, two agents.

FAQ

What knowledge base does it use?
Your global Markdown + YAML frontmatter library (default D:\knowledge) — the same one the Codex kb.cmd CLI writes.
Which tools does it add?
kb_add (typed notes, byte-compatible with the Python CLI), kb_search (title > tags > content), kb_show (full text by title/path/tag), and kb_timeline (append timestamped line after writeback).
Does it duplicate the Python CLI?
No — the heavy read side (ask/reflect/map/ingest) stays on the Python CLI; this package owns the deterministic write + quick-read surface.

Alternatives

Phant0Meow/dsh-memory-meow · KLRSL/dsh-biomemory · Spirtxiaoqi7/mindspace-dsh-local-rag

More plugins in Memory

Browse more in Memory

Guides for Memory plugins