generated: '2026-09-19'
method: probed
status: published
source: https://www.claix.dev/documentation/mcp
tools_source: https://www.claix.dev/mcp (anonymous JSON-RPC tools/list, HTTP 200, 2026-09-19)
llms_source: https://www.claix.dev/llms.txt
description: Claix ships a hosted, remote MCP server on its own domain. The documented URL is https://claix.dev/mcp;
  the apex answers 308 to https://www.claix.dev/mcp, which is the URL an MCP client actually ends up POSTing
  to. tools/list, prompts/list, resources/list and initialize all answer anonymously (tool CALLS require
  the x-api-key header or an api_key argument). The full tools/list body, with every inputSchema, outputSchema
  and annotation, is saved verbatim in claix-dev-mcp-tools-list.json.
deployment:
  mode: remote
  endpoint: https://www.claix.dev/mcp
  documented_endpoint: https://claix.dev/mcp
  install: null
  package: null
  auth: api-key
  verified: probed
  transports:
  - streamable-http
  - sse
  listings:
  - registry: smithery
    qualified_name: info-f4xz/claix
    url: https://smithery.ai/server/info-f4xz/claix
    deployment_url: https://claix--info-f4xz.run.tools
    install: npx -y @smithery/cli run info-f4xz/claix
    note: Smithery-hosted proxy of the same server; the registry record (registry.smithery.ai/servers/info-f4xz/claix,
      200) maps its APIKEY config field to the x-api-key header.
server:
  name: claix
  version_observed:
    initialize_serverInfo: 1.7.0
    docs_page: 1.9.0
    well_known_server_card: 1.5.0
  protocol_version: '2025-06-18'
  capabilities:
    tools:
      listChanged: true
    prompts:
      listChanged: true
    resources:
      listChanged: true
    completions: {}
  static_metadata: https://claix.dev/.well-known/mcp/server-card.json
auth:
  methods:
  - apikey
  apikey:
    header: x-api-key
    fallback: api_key argument on every tool (for clients that cannot send custom headers)
    note: The same key authenticates REST, A2A and MCP. No OAuth; /.well-known/oauth-* all 404.
tool_count: 17
tool_count_note: The live tools/list returned 17 tools on 2026-09-19. The docs page and llms.txt advertise
  20, the three not served being claix.spaces.create, claix.spaces.delete and claix.space_context.ask
  (documented in section 6 of the MCP page). Recorded as documented_only below rather than counted.
tools:
- name: claix.schemas.list
  category: schemas
  rest: listSchemas
  annotations:
    readOnlyHint: true
    destructiveHint: false
    idempotentHint: true
  description: List the Claix schemas available for the authenticated account. Returns id, name, type,
    schema_definition, is_agent_mode, and agent_definition. Call this first whenever you need a schema_id
    before claix.extract.* or claix.agent.* tools.
- name: claix.schemas.create
  category: schemas
  rest: createSchema
  annotations:
    readOnlyHint: false
    destructiveHint: false
    idempotentHint: false
  description: 'Create a Claix schema (POST /api/create-schema). Requires name, type (excel-json, json-excel,
    pdf-json, doc-json, img-json, txt-json) and schema_definition. Optional: is_agent_mode, agent_definition,
    resumen_agent, window_context, window_time.'
- name: claix.schemas.delete
  category: schemas
  rest: deleteSchema
  annotations:
    readOnlyHint: false
    destructiveHint: true
    idempotentHint: true
  description: Delete a Claix schema owned by the API key (POST /api/delete-schema). Requires schema_id.
- name: claix.extract.excel
  category: extraction
  rest: excelToJson
  description: Extract structured JSON from an Excel (.xlsx) or CSV file using a Claix excel-json schema.
    Processes the first sheet only. Returns typed JSON matching the schema. Use when the user attaches
    a spreadsheet or CSV.
- name: claix.extract.pdf
  category: extraction
  rest: pdfToJson
  description: Extract structured JSON from a PDF (text or scanned, max 15 MB) using a Claix pdf-json
    schema. Returns a single JSON object in data[]. Use for invoices, contracts, forms, or reports.
- name: claix.extract.doc
  category: extraction
  rest: docToJson
  description: Extract structured JSON from a Word document (.docx, .txt, .md, .rtf; max 10 MB) using
    a Claix doc-json schema. Legacy .doc is not supported.
- name: claix.extract.image
  category: extraction
  rest: imgToJson
  description: Extract structured JSON from a visible image (JPEG, PNG, WebP, HEIC/HEIF; max 15 MB) using
    a Claix img-json schema. Use for photos of documents, tickets, or labels.
- name: claix.extract.text
  category: extraction
  rest: txtToJson
  description: Extract structured JSON from already processed plain text, HTML, or XML using a Claix txt-json
    schema. Send the payload in the content field (no file). Max 300,000 characters. POST /api/txt-json.
- name: claix.convert.json_to_excel
  category: conversion
  rest: jsonToExcel
  description: Convert JSON to an Excel .xlsx file using a Claix json-excel schema. Provide json_data
    inline or file_base64 with a .json file. Returns base64-encoded xlsx on success.
- name: claix.agent.excel
  category: agent-mode
  rest: agentExcelToJson
  description: Run Excel/CSV extraction plus Agent mode semantic reasoning. Requires is_agent_mode on
    the schema. Returns data[] and agent_data with typed answers from agent_definition.
- name: claix.agent.pdf
  category: agent-mode
  rest: agentPdfToJson
  description: Run PDF extraction plus Agent mode reasoning. Requires is_agent_mode. Returns structured
    data[] and agent_data for business rules, booleans, or summaries defined in the schema.
- name: claix.agent.doc
  category: agent-mode
  rest: agentDocToJson
  description: Run document extraction plus Agent mode reasoning on .docx/.txt/.md/.rtf. Requires is_agent_mode.
    Returns data[] and agent_data.
- name: claix.agent.image
  category: agent-mode
  rest: agentImgToJson
  description: Run image extraction plus Agent mode reasoning. Requires is_agent_mode. Returns data[]
    and agent_data from visible content.
- name: claix.agent.text
  category: agent-mode
  rest: agentTxtToJson
  description: Run text/HTML/XML extraction plus Agent mode reasoning. Requires is_agent_mode. Send content
    (no file). Returns data[] and agent_data. POST /agent/txt-json.
- name: claix.window_context.get
  category: context-window
  rest: getDocument
  description: Retrieve raw content and metadata from a persisted document (GET /window-context/{document_id}).
    Requires document_id from an extraction with window_context. Returns success, document_id, file_name,
    schema_id, processed_at, and content. Free call.
- name: claix.window_context.ask
  category: context-window
  rest: documentContextAsk
  description: Ask up to 5 questions about a persisted document (POST /window-context/{document_id}).
    Requires document_id from an extraction with window_context. Each question max 400 characters. Returns
    user_ask and ia_response (null if not in the document).
- name: claix.document.delete
  category: context-window
  rest: deleteDocument
  annotations:
    readOnlyHint: false
    destructiveHint: true
    idempotentHint: true
  description: Delete a persisted document by document_id (DELETE /delete-document/{document_id}). No
    body or query params. Returns { document_id } on success. Irreversible. Free call. 404 if the document
    does not exist, is foreign, or was already purged.
documented_only:
- name: claix.spaces.create
  category: knowledge-spaces
  rest: createSpace
  note: in docs section 6
  absent from live tools/list 2026-09-19: null
- name: claix.spaces.delete
  category: knowledge-spaces
  rest: deleteSpace
  note: in docs section 6
  absent from live tools/list 2026-09-19: null
- name: claix.space_context.ask
  category: knowledge-spaces
  rest: spaceContextAsk
  note: in docs section 6
  absent from live tools/list 2026-09-19: null
prompts:
- workflow.discover-and-extract
- workflow.agent-document-analysis
- workflow.invoice-pdf
- workflow.window-context-ask
resources:
- uri: claix://docs/mcp
  name: mcp-documentation
- uri: claix://docs/openapi
  name: openapi-spec
- uri: claix://docs/tools
  name: tool-catalog
clients_documented:
- Cursor
- Claude Desktop (via mcp-remote)
- Claude Code
- Windsurf
- Cline
- Continue
- Zed
- Smithery
- Lovable
- Replit Agent
- ChatGPT connectors
- n8n
- Make
- LangGraph
x-evidence:
- url: https://www.claix.dev/mcp
  method: POST tools/list
  http_status: 200
  content_type: application/json
  fetched: '2026-09-19'
- url: https://www.claix.dev/mcp
  method: POST initialize
  http_status: 200
  serverInfo: claix 1.7.0
  fetched: '2026-09-19'
- url: https://claix.dev/mcp
  method: POST tools/list
  http_status: 308
  location: https://www.claix.dev/mcp
  fetched: '2026-09-19'
- url: https://claix.dev/.well-known/mcp/server-card.json
  http_status: 200
  content_type: application/json
  fetched: '2026-09-19'
- url: https://registry.smithery.ai/servers/info-f4xz/claix
  http_status: 200
  fetched: '2026-09-19'
descriptions_source: live tools/list (authorship probe or saved capture); filled by backfill-tools-from-authorship.py
  --descriptions, 2026-09-28