Best for
- AI coding agents needing efficient, reliable code search
- Developers using MCP clients like Claude, Cursor
P
Pare Search
Community
Structured code search operations (ripgrep + fd) as typed JSON.
- Category
- Dev Tools #295 of 438
- Stars
- ★ 139 Popular
- Transport
- stdio (local process)
- Runtime
- Node.js >= 20
- Credentials
- No credential needed
- License
- MIT
- Last commit
46FMRS · D
Pare Search provides structured output for code search, significantly reducing token usage and ideal for AI agents needing stable results. However, understand its unofficial nature and potential risks.
Strongest · Setup experience 12/20 Weakest · Reliability 6/20
Why each score
- Reliability 6/20
- Only the manifest and README were supplied; no source, tests, or CI workflow for packages/server-search/server.json were available. The MCP init handshake and tool behavior cannot be verified, and no tool-list/schema evidence exists. README claims are untrusted. The happy path is plausible but not reproducible; scored down.
- Security and permissions 8/20
- The server wraps ripgrep/fd for local read-only search, with no credential handling, external network, or destructive side effects; no red-line issues found. However, source/docs do not establish path scoping, symlink handling, or restrictions on sensitive directories; tool-filter env vars limit tool availability but not permissions. Main risks are visible but scoping is incomplete; score 8.
- Maintenance 10/20
- The repository is not archived and has an MIT license; the manifest shows v0.22.0 and the README displays CI/coverage/OpenSSF badges, but these are unverified claims. No commit dates, release history, dependency-update, or issue-response evidence was supplied. Scored at anchor 10: active appearance with governance/versioning gaps.
- Documentation 10/20
- The repo-level README covers setup, troubleshooting, client config patterns, and tool selection, and links to tool-schemas; however, server-search-specific tool parameters, output schemas, path/performance limits, and error handling are absent from the supplied material, and linked docs are not included. Hidden assumptions remain; anchor 10.
- Setup experience 12/20
- The npm package identifier and stdio transport are clear, and the README supplies npx usage, Node>=20 requirement, Windows cmd /c npx guidance, and a JSON MCP config example (for another server). No search-specific config example or execution evidence is included; stays below the static cap of 15, scored 12.
Static review · not runListed 2026-08-07
Read the FMRS scoring method →Fit and risk
What it can accessReads local filesRuns commands or code
Not for
- Full-text search of non-code files (e.g., documents, logs)
- Advanced users who need granular control over underlying ripgrep/fd parameters
Required permissions
- Read file system to search code
- Execute ripgrep and fd commands
Risks and side effects
- Search results may include sensitive information; handle outputs carefully
- Depends on local Node.js environment and network (first run)
- Server is not official; potential maintenance risk
Setup
Before you start
Runtime：Node.js >= 20
Other optional settings (1)
PARE_TOOLS
Universal tool filter in server:tool,... format that limits registered tools across all Pare servers; optional.
Install and run via npm: npx -y @paretools/search. Requires Node.js >= 20. Refer to the repository README for configuration guides to add it to MCP clients.
Check that it works
After installing (npx @paretools/search) and restarting the client, confirm the Pare Search tools appear in the client's tool list; then ask it to search for a string you know exists — a structured JSON result proves the connection works.
Troubleshooting
- Verify Node.js version >= 20
- First run may be slow; pre-cache or install globally
- Check MCP client configuration to ensure command is correct
Things to try
Once connected, you can ask your AI assistant things like:
- Search the project for all files containing parseStructuredContent
- Use ripgrep to find all TypeScript files with TODO comments and list line numbers
- Find all files named *.test.ts in this repository
- Search the src directory for the string mcpServers and show the matches
Use cases
Quickly search for patterns or filenames in large codebases
Provide structured search results to AI agents, avoiding text parsing errors
Reduce token consumption from terminal output, saving context space
Overview
Pare Search is an MCP server in the Pare project that wraps the ripgrep and fd command-line tools to provide structured, typed JSON output, replacing fragile terminal text parsing. This server focuses on code search, delivering reliable and token-efficient data for AI agents.
Similar servers
Context7
80 · B
Upstash's official server providing up-to-date third-party library docs for AI coding assistants
NPM Sentinel MCP Server
79 · B
AI-powered NPM package analysis MCP server
MartinLoop MCP Server
79 · B
An independent control layer that gives AI coding agents budgets, brakes, and receipts
GitHub MCP Server
76 · B
GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
More servers from this repository
Source revision f76328c1c7c7 Data synced 2026-10-19 Read the FMRS scoring method