Mcp Local Rag
Drives a local RAG MCP server: index, hybrid-search, and sync your docs offline
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Jul 21, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 0690453
Located SKILL.md at skills/mcp-local-rag/SKILL.md; all 4 body-referenced files returned HTTP 200. Cloned into a temp HOME and confirmed the skill dir + references copy cleanly to ~/.claude/skills/mcp-local-rag/. Could not run the RAG pipeline (needs the mcp-local-rag server, Node 22, and a model download), so I tested the Claude-side reasoning layer: for an error-stack "search my docs" task I wrote a baseline query artifact (whole stack trace dumped as the query, limit 10, keeps noise chunks) and a skill-following artifact (baseline.txt/skill.txt in scratchpad). Following the body's rules produced a materially different, tighter query ("refreshSession accessToken undefined session"), limit 5, and applied the score table to drop the 0.63 and 0.78 chunks while keeping 0.22+0.41.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 5/5
- Output vs. baseline 7/10
- Docs & honesty 5/5
What Mcp Local Rag does
Guides Claude to search, save, and maintain a local document index through the companion mcp-local-rag MCP server (or its npx CLI), covering hybrid query formulation, score-based result filtering, chunk-neighbor context expansion, PDF visual ingest, and disk-to-index sync. Triggers on phrases like "search my docs", "save this page", "read around that chunk", or "sync my index". Requires registering the mcp-local-rag server (Node 22+) with a BASE_DIR before the tools it references exist.
How to install Mcp Local Rag
git clone --depth 1 https://github.com/shinpr/mcp-local-rag.git /tmp/mcp-local-rag-src
mkdir -p ~/.claude/skills
cp -R /tmp/mcp-local-rag-src/skills/mcp-local-rag ~/.claude/skills/mcp-local-rag
# The skill only drives the mcp-local-rag MCP server — its tools (ingest_file, query_documents, sync_start...) do not exist until you register it.
# Requires Node.js 22+. Register the server (replace the path with your docs dir):
# claude mcp add local-rag --scope user --env BASE_DIR=/absolute/path/to/your/documents -- npx -y mcp-local-rag
# First index sync downloads a ~90MB embedding model; opt-in PDF visual mode downloads a local VLM under CACHE_DIR (./models/).
# CLI without MCP: npx mcp-local-rag ingest ./docs/ ; npx mcp-local-rag query "authentication API"
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Mcp Local Rag
-
/mcp-local-ragDrives a local RAG MCP server: index, hybrid-search, and sync your docs offline
It also activates on plain-language prompts like these:
-
Search my docs for last quarter's notes -
Save this page into my local RAG index -
Read around that chunk for more context
Frequently asked questions
- Is the Mcp Local Rag skill free?
- Yes. The skill itself is free from shinpr/mcp-local-rag. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Mcp Local Rag work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Tested · Works. Located SKILL.md at skills/mcp-local-rag/SKILL.md; all 4 body-referenced files returned HTTP 200. Cloned into a temp HOME and confirmed the skill dir + references copy cleanly to ~/.claude/skills/mcp-local-rag/. Could not run the RAG pipeline (needs the mcp-local-rag server, Node 22, and a model download), so I tested the Claude-side reasoning layer: for an error-stack "search my docs" task I wrote a baseline query artifact (whole stack trace dumped as the query, limit 10, keeps noise chunks) and a skill-following artifact (baseline.txt/skill.txt in scratchpad). Following the body's rules produced a materially different, tighter query ("refreshSession accessToken undefined session"), limit 5, and applied the score table to drop the 0.63 and 0.78 chunks while keeping 0.22+0.41.
- What is the Mcp Local Rag SkillProof Score?
- 8.8/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 7/10, docs & honesty 5/5.
- How do I install Mcp Local Rag?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Mcp Local Rag with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.