Best for
- AI coding assistants needing reliable, typed HTTP response data
- Developers seeking to reduce token usage and parsing errors
P
Pare HTTP
Community
Structured HTTP request operations (GET, POST, HEAD) as typed JSON.
- Category
- Dev Tools #405 of 438
- Stars
- ★ 138 Popular
- Transport
- stdio (local process)
- Runtime
- Node.js >=20
- Credentials
- No credential needed
- License
- MIT
- Last commit
32FMRS · D
Pare HTTP is an MCP server for AI agents, focusing on HTTP operations with structured JSON output, reducing token consumption and improving reliability. It's best suited for workflows that need to interact with HTTP APIs, but not for scenarios requiring alternate transports or non-HTTP functionality.
Strongest · Setup experience 9/20 Weakest · Security and permissions 4/20
Why each score
- Reliability 5/20
- The only evidence supplied is the server.json manifest and a single-sentence description in the README ('Structured HTTP request operations (GET, POST, HEAD) as typed JSON'). No source code, tool list, outputSchema, or tests for packages/server-http were provided, so the MCP init handshake and the match between declared and actual tools cannot be verified. CI and coverage badges in the README are untrusted external claims and are not counted as evidence. Under the static-review rule, key paths are not evidenced, so the score is conservatively set to 5.
- Security and permissions 4/20
- The server's functionality is issuing HTTP requests, which can target arbitrary internal or external endpoints, yet the source and docs show no evidence of SSRF protection, destination allowlisting, confirmation for dangerous operations, or credential-handling disclosure. No malware, credential theft, or covert exfiltration evidence exists, so the red-line is not triggered; however, least privilege, confirmation, and data-flow transparency are all invisible, keeping this dimension low at 4.
- Maintenance 7/20
- The repository is not archived, the license is MIT, and the manifest declares an npm package with version 0.22.0, indicating versioned release machinery. However, no commit history, release cadence, issue responses, dependency updates, or security-response channel for this server were provided; the 17 open issues have no visible disposition. Stars are discovery signals only and add no points, so maintenance is scored at 7.
- Documentation 7/20
- The manifest is a one-line description; the README is repository-level. Although it links to tool schemas, response examples, and client setup documents, none of those contents were supplied, so it cannot be confirmed that they cover packages/server-http. Specifics about tool parameters, authentication, failure modes, error handling, platform limitations, and troubleshooting are missing for this server. Documentation is usable but has clear hidden assumptions, hence 7.
- Setup experience 9/20
- The manifest reasonably implies a basic stdio install via the npm package @paretools/http (e.g., npx -y @paretools/http), and the README mentions Node.js >= 20 plus tool-filtering environment variables. But there is no concrete client configuration example for this server, no platform compatibility detail, no verification steps, and no executable CI/test evidence, so the static cap of 15 applies. The path appears simple yet under-evidenced; the score is conservatively 9.
Static review · not runListed 2026-08-07
Read the FMRS scoring method →Fit and risk
What it can accessUses the network
Not for
- Scenarios requiring non-HTTP functionality
- Running on older Node.js (<20) versions
- MCP clients that do not support stdio transport
Required permissions
- Network access to send HTTP requests
- Read/write local files (if requests involve file operations like uploads)
- System command execution (via curl)
Risks and side effects
- Potential for unintended requests if server is not secured
- Sensitive data may be exposed via HTTP requests
- Dependency on network stability
Setup
Before you start
Runtime：Node.js >=20
Other optional settings (2)
PARE_TOOLS
Universal tool filter in server:tool,comma format limiting which tools register across all Pare servers; optional.
MCP_TIMEOUT
MCP connection timeout in milliseconds for Claude Code, e.g. 30000, useful if connections time out; optional.
Install via npm and configure your MCP client, e.g., npx -y @paretools/http. Set command to npx and args to ['-y', '@paretools/http'] in client config. Ensure Node.js version >= 20.
Check that it works
After connecting the client, check the tool list for the Pare HTTP request tools, then ask it to send a GET request to https://api.github.com; receiving structured JSON confirms the install works.
Troubleshooting
- Check Node.js version meets requirement (>=20)
- Verify MCP client configuration is correct
- Ensure network connectivity is available
- Review server logs for error details
Things to try
Once connected, you can ask your AI assistant things like:
- Send a GET request to https://api.github.com and return the JSON response
- Check this URL's response status with a HEAD request
- POST this JSON payload to the test endpoint and show the response
- Restrict this server to only register the GET request tool
Use cases
AI agents performing HTTP requests and receiving structured responses
Automated API testing and debugging
Integrating HTTP calls in development workflows without manual curl parsing
Overview
Pare HTTP is one of the MCP servers in the Pare project, providing structured HTTP request operations (GET, POST, HEAD) that return typed, schema-validated JSON instead of raw terminal text. It wraps tools like curl, eliminating fragile parsing of CLI output and enabling agents to consume reliable data directly.
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 57e1980e8375 Data synced 2026-10-12 Read the FMRS scoring method