One MCP message with its redacted arguments and the API calls it made
Request
- Base URL: https://api.oveyon.com/v1
- URL: https://api.oveyon.com/v1/logs/mcp/{id}
- Auth: HTTP bearer
Path parameters
Response
The message
The MCP log entry id (GET /v1/logs/mcp/{id}).
When the request carrying this message ARRIVED (milliseconds).
One per JSON-RPC message. The API calls this message made carry the same value in mcpCallId (GET /v1/logs).
The JSON-RPC method (tools/call, tools/list, initialize…). null = no message could be read (malformed body, or a GET/DELETE answered 405).
The tool the agent called (only on tools/call).
ok; tool_error = the tool ran and returned an error — usually the error the API answered (see apiStatuses); rejected = the MCP server refused the call before running the tool (unknown tool, invalid arguments); error = the MCP request itself failed with an HTTP 4xx/5xx (malformed JSON, method not allowed, the key refused…); aborted = the client gave up before the answer. Outside tools/call, ok means the request was answered: a protocol error inside the answer (an unknown method or resource) is not read.
Why the message did not end ok at the MCP level: tool_not_found, invalid_arguments, or the error of the MCP request itself (parse_error, method_not_allowed, the authentication error…). The error of an API call the tool made is apiError (GET /v1/logs/mcp/{id}).
The HTTP status of the MCP request that carried this message.
Server-side time of that request.
How many API calls this message made (0 for protocol messages).
The HTTP status of each of those API calls, in order.
The tool (tools/call) or the resource (resources/read) the message named.
The HTTP method of the MCP request (POST; GET/DELETE are answered 405).
The JSON-RPC id the client sent (null on notifications).
The tool arguments as stored: only fields of the tool schema (others become { "omitted": "unknown_field" }), with secrets replaced by «redigido» and message bodies and attachments — file names included — (html, text, content, filename…) replaced by { "omitted": "content", "length": N }. Capped at 8 KB: arguments that do not fit the cap as valid JSON become { "omitted": "too_large", "bytes": N }, where N is the size in bytes of the arguments as received, and argumentsTruncated is true.
true when the stored arguments were cut at the cap.
Size of the MCP request body, in bytes.
Size of the tool result — this log never keeps the result itself.
The method and route pattern of the (last) API call the tool made (GET /v1/inbound/:uid/content).
The error code of that API call, on 4xx/5xx.
The caller IP as seen by our edge — for an app like ChatGPT or Claude, an address of the app provider, not of the person.
The User-Agent header (up to 255 characters).
The MCP protocol version: the one asked in initialize, or the MCP-Protocol-Version header.
The Mcp-Session-Id header, when the client sent one.
The GET /v1/logs/{id} entries of the API calls this message made.
Changes
No recorded changes to this operation across all 1 revision of this API.