Claix

Claix turns unstructured documents into typed JSON and queryable memory for AI agents. Extract PDF, Excel, Word, images, and text; ask persisted documents by document_id; reason across a knowledge space with space_id. Optional Source Tracing on extracted fields.

Skills

• Convert an Excel or CSV file to JSON using a schema
  Convert an Excel or CSV file to JSON using a schema. Accepts an .xlsx or .csv file and a schema_id, and returns transformed JSON whose keys match the schema properties exactly, plus log_id. If source verification is enabled on the schema, each field is {value, source} instead of a bare value. If the workbook has multiple sheets, only the first sheet is processed. Max file size follows the REST /excel-json limits. A2A skill id: extract-excel. REST: POST /excel-json. JSON DataPart must include "skill": "extract-excel". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. multipart/form-data equivalent: schema_id + file_base64 or file_path in the DataPart JSON. Optional space_id groups the persisted document. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extractexcelcsvjson
• Convert JSON documents to an Excel file
  Convert JSON documents to an Excel file. Accepts JSON data and a schema_id, and returns a binary .xlsx whose columns follow the schema order and names. In A2A send schema_id plus data (array or object). Aliases records and json_data are normalized to data. A2A skill id: convert-json-to-excel. REST: POST /json-excel. JSON DataPart must include "skill": "convert-json-to-excel". Required A2A parameters: skill. Body JSON fields: schema_id, data. JSON body: schema_id (required) and data (array or object). Aliases records/json_data also map to data. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  convertexceljson
• Extract structured data from a PDF using a schema
  Extract structured data from a PDF using a schema. Accepts a .pdf (selectable text or scanned) and a schema_id, and returns a single JSON object for the whole document plus log_id. Response data always contains exactly one record. If source verification is enabled on the schema, each field is {value, source}. Max file size: 15 MB. A2A skill id: extract-pdf. REST: POST /pdf-json. JSON DataPart must include "skill": "extract-pdf". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. schema_id + file_base64 or file_path in the DataPart JSON. Optional space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extractpdfjson
• Extract structured data from a text document using a schema
  Extract structured data from a text document using a schema. Accepts .docx, .txt, .md, or .rtf plus a schema_id, and returns a single JSON object for the whole document plus log_id. If source verification is enabled on the schema, each field is {value, source}. Legacy .doc (Word 97-2003) is not supported. Max file size: 10 MB; extracted text max 300,000 characters. A2A skill id: extract-doc. REST: POST /doc-json. JSON DataPart must include "skill": "extract-doc". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. schema_id + file_base64 or file_path (.docx, .txt, .md, .rtf). Optional space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extractdocxjson
• Extract structured data from an image using a schema
  Extract structured data from an image using a schema. Accepts .jpeg, .jpg, .png, .webp, .heic, or .heif plus a schema_id, and returns a single JSON object from the visible content plus log_id. If source verification is enabled on the schema, each field is {value, source}. Unreadable images return HTTP 422. Max file size: 15 MB. A2A skill id: extract-img. REST: POST /img-json. JSON DataPart must include "skill": "extract-img". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. schema_id + file_base64 or file_path (.jpeg, .png, .webp, .heic). Optional space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extractimagejson
• Extract structured data from plain text, HTML, or XML using a schema
  Extract structured data from plain text, HTML, or XML using a schema. Accepts a content string (plain text, HTML, or XML) and a txt-json schema_id. No file upload. Returns log_id. If source verification is enabled on the schema, each field is {value, source}. Max 300,000 characters. A2A skill id: extract-txt. REST: POST /txt-json. JSON DataPart must include "skill": "extract-txt". Required A2A parameters: skill, content, schema_id. Form/DataPart fields: schema_id, content, space_id. schema_id + content (plain text/HTML/XML string, not a file). Optional space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extracttexthtmlxml
• Extract structured data from spoken audio using a schema
  Extract structured data from spoken audio using a schema. Accepts .mp3, .wav, .m4a, or .ogg plus a schema_id, and returns a single JSON object from the spoken content plus log_id. If source verification is enabled on the schema, each field is {value, source}. Unintelligible audio returns HTTP 422. Max file size: 12 MB. Max duration: 10 minutes. A2A skill id: extract-audio. REST: POST /audio-json. JSON DataPart must include "skill": "extract-audio". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. schema_id + file_base64 or file_path (.mp3, .wav, .m4a, .ogg). Optional space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  extractaudiojson
• Return the raw content of a persisted document
  Return the raw content of a persisted document. Read-only. document_id is a path parameter only (no body). Returns markdown/text and metadata for a document created by an extraction with window_context enabled. Free call. A2A skill id: get-document. REST: GET /get-document/{document_id}. JSON DataPart must include "skill": "get-document". Required A2A parameters: skill, document_id. Path params: document_id. document_id is a path parameter only. No body and no questions. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  contextdocument
• Answer questions about a persisted document
  Answer questions about a persisted document. document_id goes in the URL path. questions (1–5 strings, max 400 chars each) go in the JSON body only — never embed document_id inside a question. Returns { user_ask, ia_response, log_id } aligned 1:1; ia_response items are null when the answer is not in the document. If source verification is enabled on document queries, each ia_response item is {value, source} citing markdown evidence (or source is requires_human_revision). A2A skill id: query-document. REST: POST /document-context/{document_id}. JSON DataPart must include "skill": "query-document". Required A2A parameters: skill, document_id, questions. Path params: document_id. Body JSON fields: questions. document_id goes in the URL path. questions[] of { question, format } (1–5 items; format ∈ string|int|boolean|timestamp|array; question max 400 chars) goes in the JSON body only — never inside the question text. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  contextqadocument
• Answer questions across every document in a knowledge space
  Answer questions across every document in a knowledge space. space_id goes in the URL path. questions (1–5 strings, max 400 chars each) go in the JSON body only — never embed space_id inside a question. Cross-references up to 50 active documents in the space. Returns { user_ask, ia_response, log_id }; null means no evidence in any document. If source verification is enabled on knowledge space queries, each ia_response item is {value, source} naming document_id and file_name. A2A skill id: query-space. REST: POST /space-context/{space_id}. JSON DataPart must include "skill": "query-space". Required A2A parameters: skill, space_id, questions. Path params: space_id. Body JSON fields: questions. space_id goes in the URL path. questions[] of { question, format } (1–5 items; format ∈ string|int|boolean|timestamp|array; question max 400 chars) goes in the JSON body only — never inside the question text. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  contextqaspace
• Create a knowledge space for the API key account
  Create a knowledge space for the API key account. Creates an empty knowledge space. Required JSON body field: name (max 200 chars). Returns space_id for use as optional space_id on extractions and for query-space. Free call. A2A skill id: create-space. REST: POST /create-space. JSON DataPart must include "skill": "create-space". Required A2A parameters: skill, name. Body JSON fields: name. JSON body with name (string, max 200 chars). Returns space_id. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  spaces
• Delete a knowledge space
  Delete a knowledge space. space_id is a path parameter only (no body). Deletes the space owned by the API key account. Free call. A2A skill id: delete-space. REST: DELETE /delete-space/{space_id}. JSON DataPart must include "skill": "delete-space". Required A2A parameters: skill, space_id. Path params: space_id. space_id is a path parameter only. No body. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  spaces
• Delete a persisted document
  Delete a persisted document. document_id is a path parameter only (no body). Deletes the document owned by the API key account. Free call. A2A skill id: delete-document. REST: DELETE /delete-document/{document_id}. JSON DataPart must include "skill": "delete-document". Required A2A parameters: skill, document_id. Path params: document_id. document_id is a path parameter only. No body. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  contextdocument
• Excel/CSV to JSON with Agent Mode reasoning
  Excel/CSV to JSON with Agent Mode reasoning. Same inputs as extract-excel, but the schema must have is_agent_mode enabled. Returns extraction data, agent_data from agent_definition, and log_id. If source verification is enabled on the schema, fields are {value, source}. A2A skill id: agent-extract-excel. REST: POST /agent/excel-json. JSON DataPart must include "skill": "agent-extract-excel". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. Same as extract-excel but schema must have is_agent_mode. schema_id + file_base64 or file_path. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agentexcel
• PDF to JSON with Agent Mode reasoning
  PDF to JSON with Agent Mode reasoning. Same inputs as extract-pdf, but the schema must have is_agent_mode enabled. Returns data[], typed agent_data, and log_id. If source verification is enabled on the schema, fields are {value, source}. Max PDF size: 15 MB. A2A skill id: agent-extract-pdf. REST: POST /agent/pdf-json. JSON DataPart must include "skill": "agent-extract-pdf". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. Same as extract-pdf but schema must have is_agent_mode. schema_id + file_base64 or file_path. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agentpdf
• Document to JSON with Agent Mode reasoning
  Document to JSON with Agent Mode reasoning. Same inputs as extract-doc, but the schema must have is_agent_mode enabled. Returns data[], agent_data, and log_id. If source verification is enabled on the schema, fields are {value, source}. Formats: .docx, .txt, .md, .rtf. A2A skill id: agent-extract-doc. REST: POST /agent/doc-json. JSON DataPart must include "skill": "agent-extract-doc". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. Same as extract-doc but schema must have is_agent_mode. schema_id + file_base64 or file_path. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agentdocx
• Image to JSON with Agent Mode reasoning
  Image to JSON with Agent Mode reasoning. Same inputs as extract-img, but the schema must have is_agent_mode enabled. Returns data[], agent_data, and log_id. If source verification is enabled on the schema, fields are {value, source}. Formats: JPEG, PNG, WebP, HEIC/HEIF. A2A skill id: agent-extract-img. REST: POST /agent/img-json. JSON DataPart must include "skill": "agent-extract-img". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. Same as extract-img but schema must have is_agent_mode. schema_id + file_base64 or file_path. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agentimage
• Text/HTML/XML to JSON with Agent Mode reasoning
  Text/HTML/XML to JSON with Agent Mode reasoning. Same inputs as extract-txt, but the schema must have is_agent_mode enabled. content string + schema_id; returns data[], agent_data, and log_id. If source verification is enabled on the schema, fields are {value, source}. A2A skill id: agent-extract-txt. REST: POST /agent/txt-json. JSON DataPart must include "skill": "agent-extract-txt". Required A2A parameters: skill, content, schema_id. Form/DataPart fields: schema_id, content, space_id. Same as extract-txt but schema must have is_agent_mode. schema_id + content string. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agenttext
• Audio to JSON with Agent Mode reasoning
  Audio to JSON with Agent Mode reasoning. Same inputs as extract-audio, but the schema must have is_agent_mode enabled. Returns data[], agent_data, and log_id. If source verification is enabled on the schema, fields are {value, source}. Formats: MP3, WAV, M4A, OGG. Max 12 MB / 10 minutes. A2A skill id: agent-extract-audio. REST: POST /agent/audio-json. JSON DataPart must include "skill": "agent-extract-audio". Required A2A parameters: skill, schema_id. Form/DataPart fields: schema_id, file_base64, file_path, filename, space_id. Same as extract-audio but schema must have is_agent_mode. schema_id + file_base64 or file_path. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  agentaudio
• List every schema for the API key account
  List every schema for the API key account. No parameters. Returns all schemas owned by the API key account (id, name, type, definition), newest first. Free call. A2A skill id: list-schemas. REST: GET /schemas. JSON DataPart must include "skill": "list-schemas". Required A2A parameters: skill. No parameters. Returns every schema in the account. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  schemas
• Create a schema for the API key account
  Create a schema for the API key account. JSON body: name, type (excel-json | json-excel | pdf-json | doc-json | img-json | txt-json | audio-json), and schema_definition (object). Optional is_agent_mode, agent_definition, resumen_agent, window_context, window_time, and source verification. Free call. A2A skill id: create-schema. REST: POST /create-schema. JSON DataPart must include "skill": "create-schema". Required A2A parameters: skill, name, type, schema_definition. Body JSON fields: name, type, schema_definition, is_agent_mode, agent_definition, resumen_agent, window_context, window_time, cita_por_campo. JSON body: name, type (excel-json|pdf-json|doc-json|img-json|txt-json|audio-json|json-excel), schema_definition (object). Optional is_agent_mode, agent_definition, window_context, window_time, source verification. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  schemas
• Delete a schema from the API key account
  Delete a schema from the API key account. JSON body with schema_id (UUID). Deletes the schema if it belongs to the API key account. Free call. A2A skill id: delete-schema. REST: POST /delete-schema. JSON DataPart must include "skill": "delete-schema". Required A2A parameters: skill, schema_id. Body JSON fields: schema_id. JSON body with schema_id (UUID). Query ?schema_id= also accepted by REST. Use snake_case names exactly. OpenAPI multipart `file` → file_base64 or file_path in A2A JSON.
  schemas

How to call

https://www.claix.dev/a2a

Listed in

Directories this entry was found in.