Agent700 Model Catalog Integration

Authentication: JWT or app-password bearer token on every request

Overview

OpenAPI named examples for this flow are prefixed with catalog_flow_. In ReadMe API Reference, open each endpoint below and choose the matching example from the request/response sample dropdown.

Flow diagram

sequenceDiagram
  participant App
  participant API
  App->>API: GET /api/models?category=text
  API-->>App: models[].name, capabilities, pricing
  alt New agent
    App->>API: POST /api/agents { model, organizationId, ... }
    API-->>App: agent.id
  else Existing agent
    App->>API: PUT /api/agents/{agentId} { model }
    API-->>App: new revision
  end
  App->>API: POST /api/chat { agentId, messages [, model] }
  API-->>App: response

Step 1 — List models

API Reference: Models → List available models (GET /api/models)

Query: category=text (recommended for chat agents)

OpenAPI example: catalog_flow_list_text_models (under the 200 response samples)

What to do:

1. Call the endpoint with your bearer token.
2. Iterate models and skip entries where isDeprecated is true or isActive is false.
3. Remember the chosen model's name (e.g. gpt-5.4, gemini-3.5-flash).
4. Optionally use capability fields (supportsMcp, supportedReasoningEfforts, supportsExtendedThinking, pricingTiers) to filter or configure later requests.

Step 2 — Set the model on an agent

You need a saved agent before chat. Pick one of the following.

Option A — Create a new agent

API Reference: Agents → Create agent (POST /api/agents)

OpenAPI example: catalog_flow_create_agent

Request highlights:

• organizationId (required)
• model — catalog name from step 1
• name, masterPrompt, and other fields as needed

What to do: Save agent.id from the response; use it as agentId in step 3.

Option B — Update an existing agent

API Reference: Agents → Update agent (new revision) (PUT /api/agents/{agentId})

OpenAPI example: catalog_flow_update_agent_model

Request highlights:

• Only { "model": "<catalog name>" } is required for a model-only change.
• Other revision fields are preserved from the current revision.

Step 3 — Chat

API Reference: Chat → Chat completion (POST /api/chat)

Always include agentId for this integration pattern.

Option A — Use the agent's configured model

OpenAPI example: catalog_flow_chat_with_agent

Omit model on the chat request; the agent revision from step 2 supplies it.

{
  "agentId": "<uuid from step 2>",
  "messages": [
    { "role": "user", "content": "Hello" }
  ]
}

Option B — One-off model override

OpenAPI example: catalog_flow_chat_model_override

Use when you want to try another catalog model without creating a new agent revision.

{
  "agentId": "<uuid>",
  "model": "<catalog name from step 1>",
  "messages": [
    { "role": "user", "content": "Your message" }
  ]
}

Example index (OpenAPI / ReadMe)

Search API Reference for "Catalog flow" to jump to these samples quickly.

Capability-aware chat options

After step 1, use the catalog entry for the model you selected:

See the Chat completion operation description and CHAT.md for full parameter lists.

Common errors