MCP Server Auditor: lint, probe, compare
Pricing
Pay per event + usage
MCP Server Auditor: lint, probe, compare
Audit any MCP server before you ship it or pay for it: read its real tool list, probe keyless endpoints, lint the TypeScript source, compare two servers and diff tool schemas. Keyless MCP server for Claude, ChatGPT, Cursor and any MCP client. Pay per tool call only.
Pricing
Pay per event + usage
Rating
0.0
(0)
Developer
a e
Maintained by CommunityActor stats
0
Bookmarked
1
Total users
0
Monthly active users
2 days ago
Last modified
Categories
Share
mcp-auditor-mcp
Audit MCP servers and the APIs behind them - no API key.
Served over the Model Context Protocol so any MCP client (Claude, Cursor, ChatGPT connectors, custom agents) can use it. Keyless: no third-party account or API key needed.
Connect
Standby MCP endpoint (streamable HTTP):
https://mrfentmen--mcp-auditor-mcp.apify.actor/mcp
Authenticate with your Apify API token:
Authorization: Bearer <YOUR_APIFY_API_TOKEN>
Tools
probe_keyless
Ask an API whether it answers with no credentials, and report its status, response shape and rate-limit headers.
audit_endpoint
Handshake with a public MCP endpoint, read its tool list and report tool count, naming, descriptions, annotations and parameter gaps.
audit_tools
Same audit as audit_endpoint, but run on a tools/list payload you paste in. Use it for endpoints behind credentials or a draft before deploy.
compare_servers
Measure how much two MCP servers overlap, so near-duplicate servers are obvious before they compete. Each side is an endpoint URL or a pasted tools/list payload.
check_listing
Validate Store copy against the real limits and the real category list, so a push cannot fail on a field length or a made-up category.
lint_source
Lint pasted MCP server TypeScript: tool count, duplicate names, tool and parameter descriptions, annotations, missing try/catch, fetch calls with no timeout, type suppressions, placeholders and dead code.
generate_tool_registration
Turn a tool spec into a complete server.ts in house style: imports, registerTool blocks, zod schemas and try/catch, ready to paste.
diff_tool_schema
Compare two tools/list replies and separate breaking changes from safe additions, so a release cannot silently break callers.
registry_lookup
Search the official MCP registry for a server name or keyword and see whether the name is taken, who published it, which versions exist and what transports they use.
npm_package_check
Check an npm package before a registry submission: is the name free, how old is the last release, how many downloads, and does the manifest carry mcpName, a bin, keywords and a license.
call_tool
Handshake with a public MCP endpoint, confirm the tool is in its list, then actually call it and report the latency, the content blocks and whether it returned a tool-level error.
server_json_check
Validate a server.json against the rules the official registry enforces: name shape and namespace, description length, semver, repository, packages or remotes, transports, and secret environment variables.
Pricing
Pay per event. You only pay for what you use:
| Event | Price (USD) | When |
|---|---|---|
| Tool call | $0.01 | Each MCP tool call |
| List request | $0.001 | Listing tools/resources/prompts |
| Resource request | $0.001 | Reading an MCP resource |
| Prompt request | $0.001 | Running an MCP prompt |
| Completion request | $0.0001 | MCP completion request |
Data source
Public read-only API. This Actor never writes to your account and never asks for third-party credentials.
Source
Built from the ts-mcp-proxy template.