igeo.ai developer documentation
Connect an AI assistant or your own integration to your igeo.ai workspace. Read tracked prompts, topics, competitor data, AI mentions, sentiment, recommendations, and supported sources. Supported write operations let authorized integrations manage monitoring data and work with generated content.
Choose an interface
| Interface | Use it for | Discovery |
|---|---|---|
| MCP | An assistant that discovers and calls workspace tools | tools/list, allowed_routes, and MCP resources |
| Public REST API | Direct HTTP integrations and reporting | Public API OpenAPI schema |
| OpenAPI Actions | A client importing the read-only connector schema | Actions OpenAPI schema |
The production API base URL is https://api.igeo.ai. The MCP endpoint is https://api.igeo.ai/mcp. Operations are limited to the public allowlist and the authenticated workspace. The full application API includes private routes that are outside this integration surface.
Get an API key
Sign in and open MCP & API keys. Active workspace members can manage keys. Copy the new key when it is displayed and record the workspace ID from the generated configuration. The secret is shown only when the key is created.
Store the key in a server-side secret store or local environment variable. Send these headers on authenticated requests:
Authorization: Bearer <YOUR_API_KEY>
X-Workspace-Id: <YOUR_WORKSPACE_ID>
Use a key scoped to the required operations. For example, a reporting integration needs read. Public REST writes require an owner, admin, or editor role as well as the appropriate scope. MCP tools also require an owner, admin, or editor role. See authentication and permissions before enabling mutations.
Make a first read
Set IGEO_API_KEY and IGEO_WORKSPACE_ID privately in your environment, then request the workspace's account details:
curl --fail-with-body \
"https://api.igeo.ai/accounts/$IGEO_WORKSPACE_ID" \
--header "Authorization: Bearer $IGEO_API_KEY" \
--header "X-Workspace-Id: $IGEO_WORKSPACE_ID"
For an assistant connection, follow the MCP setup guide. After initialization, discover the tool schemas, call allowed_routes, then use a named read tool such as prompt_work_priorities. Named tools use the workspace supplied by authentication; do not invent another account ID.
Build from observed workspace data
Fetch the current state before proposing changes. Keep the returned record IDs, date ranges, and pagination metadata with your results. Read the tool or OpenAPI schema before each new operation and check both HTTP and application-level errors.
Review proposed content or monitoring changes before applying them. If a write times out, first inspect the affected records or generation status before deciding to repeat it. These guides do not specify a universal idempotency-key contract.
Continue
- MCP: connect, discover, and run a read workflow
- Authentication: keys, OAuth, permissions, and scopes
- REST API: schemas, requests, pagination, and errors
- Product MCP documentation
- Product Public API documentation and data-use terms
JavaScript SDK and CLI
The dependency-free JavaScript client includes TypeScript declarations and a read-only CLI for Node.js 22 or newer. Download the igeo.ai SDK and CLI release and install the public artifact:
npm install https://www.igeo.ai/downloads/igeo-ai-sdk-0.1.0.tgz
npx --no-install igeo info
npx --no-install igeo schema --summary
The installed package name is @igeo-ai/sdk. This download is available directly from igeo.ai; npm registry publication is a separate release step. The artifact's SHA-256 is b34efe4a0c551472f5efc0158af39ff6d6449fd05d007b7fe47387dccb5f8551.
Store IGEO_API_KEY and IGEO_WORKSPACE_ID privately before an authenticated read. The SDK makes one request per call, refuses redirects, and does not automatically retry mutations. The CLI can read public schema metadata and authorized workspace records; it cannot write workspace data. Read the package README for cancellation, typed errors, query filters, and client configuration.
Python SDK
The Python 3.10+ client uses the standard library and the same workspace authentication. Install the Python wheel directly:
python -m pip install "https://www.igeo.ai/downloads/igeo_ai-0.1.0-py3-none-any.whl#sha256=47a58ea4c5ddf34f39933df496c27abfa85de02a0144ed2a04c9f2527710bb11"
Import the client from igeo_ai. It supports typed errors, query filters, per-request timeouts, and anonymous schema retrieval. It refuses redirects and makes no automatic retries. The source archive includes the usage guide and license. These are direct igeo.ai downloads; a PyPI listing is a separate publication step.
Endpoint availability, input fields, and allowed operations should be checked against the connected server and its published schemas.