API and agents
Use a CSV Analytics API with AI Agents
A dashboard is useful for a person at a screen. An API makes the same analysis available to an authorized workflow, so an agent can submit data, request a structured result, and return a concise answer to the operator.
Upload
Bring the CSV you already use.
Understand
Start with the brief and evidence.
Act
Ask, edit, filter, and export.
Use scoped tokens, not provider secrets
An external agent should receive an I.S.A.A.C API token with the smallest practical scope. Groq, database, payment, and deployment secrets stay on the server and are never placed in an agent prompt or browser bundle.
Return evidence with the answer
A useful agent response includes the dataset identity, relevant columns, filters, key values, and uncertainty. The agent should link back to the dashboard or export rather than returning an unsupported conclusion.
Build the connector around HTTP first
Claude, Codex, n8n, Dify, and custom agents can call documented HTTPS endpoints. An MCP server or native connector can sit on top of that contract later, but it should not be described as available until it is published and tested.
Common questions
Can an agent access the Groq API key?
No. The agent uses an I.S.A.A.C token and the server handles the model call; provider secrets remain server-side.
Is this already a native MCP plugin?
The token-authenticated API is the foundation for agent integrations. A native MCP package should be treated as a separate release with its own documentation and security review.
Keep exploring
See what your own CSV is saying.
Try the browser workspace with a real file and keep the public guide open when you need help.
Start free