McpToolset.
Capabilities
- Create a Word, PowerPoint, Excel or PDF file, or an artifact, from a request or from Markdown the agent wrote.
- Edit a file: only what is asked changes, and
get_reportlists every change. - Convert a file to another format with no AI, for example any file to Markdown so the agent can read it.
- Read what people saved in an artifact, such as poll votes.
Setup
This integration uses a Doconda API key inDOCONDA_API_KEY (from API keys). Gemini reads
GOOGLE_API_KEY, or uses Vertex AI with GOOGLE_GENAI_USE_VERTEXAI=TRUE.
With the remote server
Indocuments/agent.py:
documents/__init__.py with from . import agent, then run adk web (or adk run documents) and ask, for
example, “make an Excel file with the event budget”. The agent answers with the spreadsheet’s download links.
With the local server
timeout=300 matters locally: it is also the wait per call (5 s by default), and a tool call can take up to about 4
minutes. A document that takes longer comes back with its id and status; the agent calls get_document later.
Tools
The agent gets the eight tools of the MCP server; see the tools for what each one does. A failed call returns{ "error": "<code>", "message": "…" }, so the agent can correct itself.
Troubleshooting
- The local call times out — The default 5 s per call is shorter than a document takes. Set
timeout=300onStdioConnectionParams. 401from the remote server —DOCONDA_API_KEYis missing or invalid, so theAuthorizationheader is empty or wrong.- The download link stopped working — File links expire after 5 minutes; the agent calls
get_documentfor fresh ones.
Useful links
- MCP server: local and remote setup.
- Google Vertex AI and Gemini: the SDK tools with function calling.
- doconda-examples on GitHub: ready-to-run scripts.
- API reference.