Last updated

iblai-api-integration

Mirrored from iblai/base · skills/organizations/iblai-api-integration/SKILL.md. This page is generated — edit it in the repository, not here.

Manage an organization's account-level integrations via the API: connect the organization to external LLM providers, store the keys of services it uses (integration credentials: Stripe, Google Drive, Dropbox, Vercel, …), and issue Platform API tokens. Use when wiring an ibl.ai organization up to model providers or services, or issuing API keys.

Auth & conventions

  • Base URL: https://api.iblai.app
  • Header: Authorization: Api-Token $IBLAI_API_KEY on every request.
  • Path vars: {org} = $IBLAI_ORG, {username} = $IBLAI_USERNAME.
  • Not connected yet? Run /iblai-api-login first to populate IBLAI_ORG, IBLAI_USERNAME, and IBLAI_API_KEY.

Reads

LLMs

Third-party model provider keys. DM base: /api/ai-account/….

  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/masked-llm-credential/ — list configured providers (masked).
  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/credential/schema/ — provider field schema.
  • GET https://api.iblai.app/dm/api/ai-mentor/orgs/{org}/users/{username}/mentor/llms/ — available provider list.
  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/users/{username}/default-llm-key-usage — whether the organization may borrow ibl.ai’s LLM keys: {"use_main_llm_key": true} (or false) once the switch is stored, {"use_main_credentials": false} when it never was. It reports the stored switch only, which ibl.ai sets; a free trial or the deployment default can still let the organization borrow.

Services

Integration credentials: the keys of other services the organization uses. DM base: /api/ai-account/….

  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/integration-credential/[?name=] — list credentials as [{ id, name, value, platform }]. An admin gets every credential with its sensitive fields masked; ?unmask_sensitive_fields=true returns them in full to an admin of {org}, for the organization's own credentials only (see /iblai-vibe-credential for the token grant). Anyone else gets only the non-sensitive fields.
  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/masked-integration-credential/ — list configured services (masked).
  • GET https://api.iblai.app/dm/api/ai-account/orgs/{org}/integration-credential/schema/v2/[?name=] — each service's fields with is_sensitive (and required: false on an optional one).

APIs

Platform API keys. Same endpoints used to mint tokens elsewhere — see /iblai-api-token for the full reference.

  • GET https://api.iblai.app/dm/api/core/platform/api-tokens/?platform_key={org} — list tokens.

Writes

LLMs

Third-party model provider keys. DM base: /api/ai-account/….

  • POST https://api.iblai.app/dm/api/ai-account/orgs/{org}/credential/ — add an LLM provider:
    {
      "name": "provider, e.g. azure_openai (required)",
      "value": "object — provider fields (required)",
      "platform": "{org} (required)"
    }
  • PATCH https://api.iblai.app/dm/api/ai-account/orgs/{org}/credential/ — replace a provider's key, same body as the POST. For every provider but azure_openai the new value replaces the stored one, so send every field; azure_openai deep-merges the models you send into the stored ones.
  • DELETE https://api.iblai.app/dm/api/ai-account/orgs/{org}/credential/ — remove an LLM provider. Destructive — confirm with the user first.
    {
      "name": "string (required)",
      "platform": "string (required)",
      "model_name": "string (optional, single Azure model)"
    }

Services

Integration credentials. DM base: /api/ai-account/….

  • POST https://api.iblai.app/dm/api/ai-account/orgs/{org}/integration-credential/ — add a service's key, with every field of its schema (an optional one left blank as ""):
    {
      "name": "provider (required)",
      "value": "object (required)",
      "platform": "{org} (required)"
    }
  • PATCH https://api.iblai.app/dm/api/ai-account/orgs/{org}/integration-credential/ — change a service's key, { "name", "platform", "value" } with the fields to change in value (the SDK sends only those).
  • DELETE https://api.iblai.app/dm/api/ai-account/orgs/{org}/integration-credential/ — remove a service's key. Destructive — confirm with the user first; a VM environment variable that reads it stops working.
    {
      "platform": "string (required)",
      "name": "string (required)"
    }

APIs

Platform API keys. Same endpoints used to mint tokens elsewhere — see /iblai-api-token for the full reference.

  • POST https://api.iblai.app/dm/api/core/platform/api-tokens/ — create a token:
    {
      "username": "string",
      "name": "string",
      "key": "",
      "platform_key": "string",
      "created": "string",
      "expires": "string",
      "expires_in": "string"
    }
  • DELETE https://api.iblai.app/dm/api/core/platform/api-tokens/{name}?platform_key={org} — revoke a token. Destructive — confirm with the user first.

Example

Connect an Azure OpenAI provider to the organization:

curl -X POST \
  "https://api.iblai.app/dm/api/ai-account/orgs/$IBLAI_ORG/credential/" \
  -H "Authorization: Api-Token $IBLAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "azure_openai",
    "value": {
      "api_key": "sk-...",
      "api_base": "https://my-resource.openai.azure.com",
      "api_version": "2024-02-15-preview"
    },
    "platform": "'"$IBLAI_ORG"'"
  }'

Notes

  • LLMs and services are account-level — they apply to the whole organization, not a single agent.
  • Adding a provider the organization already holds is a 400 with non_field_errors; replace its key with the PATCH instead. azure_openai is the exception: a POST adds its models to the stored ones.
  • A service's key can also be used by an agent's virtual machine: create a virtual-machine-secret/ with source_credential_id (the credential's id) and source_field (its one sensitive field, key for Stripe), and the VM reads the key without ever seeing it. A service signed in with client_id / client_secret has no such field; only ibl.ai uses its key.
  • List endpoints return masked credentials; fetch the matching schema endpoint to learn which fields a provider expects before POSTing.
  • The API-token endpoints are shared with /iblai-api-token; tokens created via either skill are the same Platform API tokens.
  • LLM credential names and what each holds: openai (OpenAI key), anthropic (Anthropic key), gemini_google_api_key (Gemini API key), google (Google Cloud service account, Vertex AI), bedrock (AWS Bedrock), gateway_azure_foundry (Azure AI Foundry), iblai (ibl.ai’s built-in access); groq, xai, deepseek, perplexity and nvidia hold that provider’s key. A model uses the organization’s own key first, then ibl.ai’s if the organization may borrow it (default-llm-key-usage above).
  • Service drive (Google Drive) holds client_id and developer_key. client_id must be the full Google OAuth client id, <project-number>-<x>.apps.googleusercontent.com, with the drive.file scope: the leading number is sent to Google as the picker's appId.

Copyright © ibl.ai | support@iblai.zendesk.com