enquire-mcp

Community

Your Obsidian vault becomes the private memory and document intelligence layer for every AI agent.

How to install ↓ Compare GitHub source repository ↗

Category: Other #25 of 230
Stars: ★ 34 Popular
Transport: stdio (local process)
Runtime: Node.js
Credentials: No credential needed
License: MIT
Last commit: 5 days ago
Tools: 46

67FMRS · C

enquire-mcp is a feature-rich Obsidian MCP server prioritizing local-first and privacy, offering powerful hybrid retrieval and RAG. It's suitable for professionals and privacy-conscious users. Note its extensive features require configuration and include experimental elements; consider waiting for stable release.

Strongest · Security and permissions 16/20 Weakest · Reliability 10/20

Reliability

10/20

Security and permissions

16/20

Maintenance

12/20

Documentation

15/20

Setup experience

14/20

Why each score

Reliability 10/20: The server manifest and README provide clear startup arguments (serve --vault) and consistent tool count (46) and MCP protocol version (1.29). However, the manifest requires a positional 'subcommand' but examples use 'npx ... serve --vault'; package name and version match. Static review cannot verify tool behavior matches descriptions, and no actual test files or CI details are fully visible. By calibration, reliability cannot exceed 12 without execution evidence. Clear startup path and expected handshake, but lack of verifiable execution evidence, so 10.
Security and permissions 16/20: Read-only by default, write tools require explicit --enable-write, realpath checks reject symlink escapes, HTTP transport requires bearer auth with constant-time comparison and rate limiting, and privacy filters are fail-closed. These are strong security practices. No red flags for malware or overreach. Deductions for lack of code-level verification and remote transport requiring user-configured tunnels. Score 16, no red lines.
Maintenance 12/20: Repository has MIT license, release process (npm + GitHub), CI badges, clear semver versioning. Static review cannot confirm commit frequency or issue response times, and claimed 1720 tests are not verifiable here. Clear maintainer and security contact. Deduct for unverifiable governance aspects, score 12.
Documentation 15/20: README is extremely detailed: install, config for multiple clients, security model, FAQ, performance benchmarks, comparison table, API docs links. Clear privacy disclosure and limitations (languages, performance). Static review sees high quality, but some claims (e.g., '#1', 1720 tests) are marketing and unverified. Deduct for unverified evidence, score 15.
Setup experience 14/20: Install command is clear (npm install -g), provides config examples for major clients including VS Code review-and-install URI and ChatGPT remote HTTP config, plus a configure command to auto-generate config. Basic TF-IDF mode works immediately, but full hybrid requires model download. Complexity acceptable but requires Node.js and npx, and local model download may affect experience. Evidence is strong, score 14 (within static cap of 15).

Static review · not runListed 2026-09-08

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesUses the network

Best for

Privacy-conscious users who want local AI memory.
Users with multiple AI clients needing a unified memory layer.
Obsidian enthusiasts who want to keep Markdown portable.

Not for

Users requiring cloud sync or remote access without setup (unless using HTTP transport with proxy).
Those wanting zero technical configuration or model download.
Users needing only a simple global search without AI integration.

Required permissions

Read access to the Obsidian vault directory (required).
Create and edit notes (only when `--enable-write` is set).
Download models and language packs (only via setup, build-embeddings, install-model, install-ocr-lang commands).

Risks and side effects

Prompt injection: Vault content may contain instructions, so prompt must state content is data, not instructions.
Write operations: If write enabled, accidental modification possible; use dry_run and version control.
Local model download: First-run downloads (approx 118MB) require network.
Path traversal: Ensure vault path is correct and clients cannot access outside vault; server has validation.

Setup

Before you start

Runtime：Node.js

Install globally: npm install -g @oomkapwn/enquire-mcp.
Add MCP config to your client (see docs or use enquire-mcp configure).
Example config in install_config.

claude_desktop_config.json

{
  "mcpServers": {
    "obsidian": {
      "command": "npx",
      "args": ["-y", "@oomkapwn/enquire-mcp", "serve", "--vault", "/path/to/vault"]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "obsidian": {
      "command": "npx",
      "args": [
        "-y",
        "@oomkapwn/enquire-mcp",
        "serve",
        "--vault",
        "/path/to/vault"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add obsidian -- npx -y @oomkapwn/enquire-mcp serve --vault /path/to/vault

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Run `enquire-mcp doctor --vault <path>` for a health check; the obsidian_* tools (e.g. obsidian_search, obsidian_stats) should appear in the client's tool list, and an obsidian_search call should return a cited note path, confirming the connection works.

Troubleshooting

1. Verify vault path exists.
2. Check MCP config uses absolute paths.
3. Run `enquire-mcp doctor` for readiness.
4. Ensure firewall allows local process (for stdio).
5. Check version compatibility; README and CHANGELOG indicate potential issues with Claude Desktop.

Things to try

Once connected, you can ask your AI assistant things like:

Find every note where I discussed pricing strategy and summarize the evolution.
What was my decision on PostgreSQL vs MongoDB? Cite the daily note.
What pages of the LLaMA-3 paper PDF talk about scaling?
Show me topical communities in my research vault.

Tools 46

obsidian_search read-only

Performs hybrid retrieval (RRF-fused) returning cited notes and PDFs.

obsidian_hyde_search read-only

Performs HyDE-augmented search for better semantic recall.

obsidian_search_text read-only

Searches notes by textual content.

obsidian_full_text_search read-only

Full-text search using FTS5.

obsidian_semantic_search read-only

Semantic search based on embeddings.

obsidian_embeddings_search read-only

Direct embedding vector search.

obsidian_find_similar read-only

Finds similar notes to a given note.

obsidian_resolve_wikilink read-only

Resolves a wikilink to its target note.

Show 38 more tools

obsidian_get_backlinks read-only

Gets backlinks pointing to a note.

obsidian_get_outbound_links read-only

Gets outbound links from a note.

obsidian_get_note_neighbors read-only

Gets neighbors of a note (directly linked notes).

obsidian_get_unresolved_wikilinks read-only

Gets unresolved wikilinks in a note.

obsidian_find_path read-only

Finds a path between two notes in the graph.

obsidian_get_communities read-only

Detects topical communities in notes (GraphRAG-light).

obsidian_frontmatter_get read-only

Gets frontmatter of a note.

obsidian_frontmatter_search read-only

Searches notes by frontmatter fields.

obsidian_dataview_query read-only

Executes a Dataview query.

obsidian_list_tags read-only

Lists all tags across notes.

obsidian_read_note read-only

Reads the content of a note.

obsidian_list_notes read-only

Lists notes in the vault.

obsidian_get_recent_edits read-only

Gets recently edited notes.

obsidian_stale_notes read-only

Lists notes that are stale or outdated.

obsidian_open_questions read-only

Lists open questions or todos.

obsidian_context_pack read-only

Generates a context pack of related notes.

obsidian_chat_thread_read read-only

Reads a chat thread file.

obsidian_open_in_ui read-only

Opens a note in the Obsidian UI.

obsidian_stats read-only

Gets statistics about the vault.

obsidian_read_pdf read-only

Extracts text from a PDF, with page citations.

obsidian_list_pdfs read-only

Lists PDFs in the vault.

obsidian_ocr_pdf read-only

Performs OCR on a PDF page.

obsidian_read_canvas read-only

Reads a Canvas file's content.

obsidian_list_canvases read-only

Lists Canvas files in the vault.

obsidian_list_bases read-only

Lists Bases (databases) in the vault.

obsidian_read_base read-only

Reads a Base's content.

obsidian_query_base read-only

Queries a Base.

obsidian_create_note writes

Creates a new note.

obsidian_append_to_note writes

Appends content to a note.

obsidian_rename_note writes

Renames a note.

obsidian_replace_in_notes writes

Replaces text in notes.

obsidian_archive_note writes

Archives a note.

obsidian_frontmatter_set writes

Sets frontmatter of a note.

obsidian_chat_thread_append writes

Appends to a chat thread.

obsidian_lint_wiki read-only

Checks for wikilink issues.

obsidian_paper_audit read-only

Audits papers or documents for completeness.

obsidian_validate_note_proposal read-only

Validates a note proposal.

obsidian_mark_useful writes

Marks retrieved results as useful for feedback loop.

Use cases

Provide long-term memory for AI agents across sessions, preserving project history, decisions, and knowledge.

Serve as a personal second brain with multilingual, multi-format semantic search.

Enable agentic RAG and context engineering with interpretable scores and citations.

Supported clients

Claude DesktopPartial support

Claude Code

CursorPartial support

VS CodePartial support

ChatGPTPartial support

CodexPartial support

OpenClawPartial support

Listed from the project's documentation, not tested by this site.

Overview

enquire-mcp is a local-first MCP server that turns your existing Obsidian vault (including Markdown, PDFs, Canvas, and Bases) into a persistent, queryable knowledge database for any MCP-compatible AI agent. It features hybrid retrieval (BM25, TF-IDF, multilingual embeddings), local BGE reranking, HNSW indexing, freshness-aware search, and cited sources. It's read-only by default with zero cloud calls during serve, and supports 50+ languages.

Similar servers

Hevy MCP 77 · B

Manage your Hevy workout data from AI assistants via MCP.

★ 517 · Tools 26 Compare with this →

Firecrawl MCP Server 75 · B

Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents

★ 7.6k · Tools 11 Compare with this →

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 659 · Tools 11 Compare with this →

Compartment 75 · B

Encrypted, fully offline long-term memory for AI agents

★ 583 · Tools 14 Compare with this →

Compare with other MCP servers

Source revision dc3808b79b72 Data synced 2026-10-19 Read the FMRS scoring method