YYTbit/dsh-plugin-rag
Local knowledge base RAG for DeepSeek Harness
A local knowledge-base RAG plugin: it indexes project files and exposes keyword search so the agent can pull relevant context without an embedding service. The README lists what is indexed — Markdown (.md/.txt), code (.ts/.js/.py/.rs/.go/.java/.c/.cpp), config (.json/.yaml/.toml/.env) and docs (.html/.xml/.sql/.graphql) — and what is skipped (node_modules, .git, dist, build, pycache). The flow is: scan directories for text files, build a TF-IDF keyword index, inject index stats into the system prompt, and let the agent call the rag CLI when it needs a lookup. maxResults and maxBytes bound how much is fed back; enableInjection toggles the prompt-side stats.
Install
dsh plugin --profile your-profile add dsh-plugin-ragnpm package dsh-plugin-rag 0.1.1 (registry-verified 2026-09-12; repository field → github.com/YYTbit/dsh-plugin-rag). README install line quoted verbatim — substitute your profile name for your-profile. Optional config row (- id: rag / name: dsh-plugin-rag) accepts indexPaths, maxResults, maxBytes and enableInjection. A standalone rag CLI ships with the package: rag "authentication" ./src searches, rag --index ./src ./docs rebuilds the index.
Compatibility
DeepSeek Harness (dsh) with the standard plugin manager. Pure local TF-IDF search — the README states no external APIs, embedding models or heavy dependencies are required.
Details
- Repo: YYTbit/dsh-plugin-rag
- Category: Files & Data
- Stars: 1
- Version: npm dsh-plugin-rag 0.1.1 (registry-verified 2026-09-12)
- Last push: 2026-08-14
- First seen: 2026-08-13
Recent updates
The README documents behaviour rather than a release history: the search CLI (rag), the indexer's file-type allow-list and skip list, and the four config keys. There is no published changelog or migration section. Version 0.1.1 on the public registry (verified 2026-09-12).
FAQ
- How do I install dsh-plugin-rag?
- Run dsh plugin --profile <your-profile> add dsh-plugin-rag — the README's install line with your profile name substituted.
- Does it need an embedding model or API key?
- No. The README states it is pure local TF-IDF search with no external APIs or embedding models; you can also use the bundled rag CLI directly outside a session.
- Which files are indexed?
- Markdown, common source-code extensions, config formats and doc formats (see the README's lists); node_modules, .git, dist, build and pycache are skipped automatically.
Alternatives
ZSeven-W/dsh-openpencil · Tabbit-Browser/dsh-plugin · HuanLinOTO/dsh-plugin-mineru