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). The model provider
reads its own key, for example ANTHROPIC_API_KEY for @ai-sdk/anthropic.
With the SDK tools
text is the model’s final answer, with the presentation’s download links. isStepCount(5) caps the loop at 5 steps,
enough to create a document and answer. The same tools work with streamText and in a Next.js route: the agent calls
Doconda on your server, and the API key never reaches the browser.
With the MCP client
Tools
The agent gets the eight Doconda tools; see the tools for what each one does.doconda.tools() returns them with name, description, inputSchema (JSON Schema, wrapped with jsonSchema) and
run. A failed call returns { "error": "<code>", "message": "…" } instead of throwing, so the model can correct
itself.
Troubleshooting
Doconda: pass { apiKey } or set DOCONDA_API_KEY.— The SDK found no key: setDOCONDA_API_KEY, or pass{ apiKey }toDoconda.textis empty — The loop hit the step limit before the model’s last answer. Raise the count inisStepCount.- A tool result has
"error": "insufficient_balance"— The project’s balance can’t cover the document: top up in the dashboard, or ask forquality: "fast". - The download link stopped working — File links expire after 5 minutes; the agent calls
get_documentfor fresh ones.
Useful links
- TypeScript SDK.
- MCP server: local and remote setup.
- doconda-examples on GitHub: ready-to-run scripts.
- API reference.