paraphrase
Paraphrase Text
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized paraphrasing API.
Ask your agent in plain language. Each of these maps to a real tool on the server.
paraphrase
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized paraphrasing API.
text_completion
DEPRECATED — use `chat_completion` instead. AI21 retired the Jurassic-2 text completion API.
segment_text
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized text segmentation API.
get_file
Retrieve metadata for a specific file in your AI21 document library by its ID.
maestro_run
Create a Maestro AI agent run with model selection, compute budget, validation requirements, and optional file or web search. Runs may be asynchronous; use the returned runId with get_maestro_run until the run completes.
text_improvements
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized text improvements API.
summarize
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized summarization API.
contextual_answer
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized contextual answers API.
grammar_check
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized grammar correction API.
list_files
List files in your AI21 document library. Optionally filter by labels. Returns file metadata including name, type, size, status, and labels.
delete_file
Permanently delete a file from your AI21 document library.
update_file
Update metadata of a file in your AI21 document library. You can modify labels and the public URL.
conversational_rag
Ask questions over your uploaded documents using conversational RAG. Supports multi-turn conversations, document filtering by labels or file IDs, and configurable retrieval strategies. The system determines whether to answer from model knowledge or retrieved documents.
chat_completion
Generate a chat completion using AI21's Jamba models. Supports multi-turn conversations with system, user, assistant, and tool messages. Optionally enable JSON mode, pass documents for context, or define tools for function calling.
summarize_by_segment
DEPRECATED — use `chat_completion` instead. AI21 retired the specialized segment summarization API.
upload_file
Upload a TXT, Markdown, HTML, PDF, or DOCX document to the AI21 file library for retrieval. Supply UTF-8 text or base64-encoded file bytes. Processing is asynchronous; use get_file to check status before querying it.
download_file
Prepare a library file for download and return its metadata. Use list_files or upload_file to obtain the file ID.
get_maestro_run
Retrieve a Maestro run by its runId to check asynchronous status, generated output, validation results, and failure details.
Every client uses the same Magic MCP URL, and so does any other MCP client. Add it once and sign in with SSO.
{"mcpServers": {"metorial": {"url": "<your-magic-mcp-url>"}}}
claude mcp add --transport http metorial <your-magic-mcp-url>
codex mcp add metorial --url <your-magic-mcp-url>codex mcp login metorial
One governed layer for every app your agents reach, not a separate connector per tool.
Generate Jamba chat responses, create and monitor Maestro runs, query documents with conversational retrieval, and upload, list, update, download, or delete library files. Retired Jurassic-2 and specialized text APIs remain available as deprecated compatibility tools with migration guidance.
Common questions about connecting Studio by Ai 21 Labs to AI agents with Metorial.