API reference

Catalog

Catalog routes expose application-owned static definitions. They do not include workspace configuration, credentials, enablement, or effective availability.

Catalog calls

Method	Path	Description
`GET`	`/api/v1/catalog/providers`	Lists supported static provider types.
`GET`	`/api/v1/catalog/models`	Lists active static logical model definitions.
`GET`	`/api/v1/catalog/personas`	Lists packaged built-in Persona definitions.
`GET`	`/api/v1/catalog/skills`	Lists packaged built-in Skill definitions.

List provider catalog

GET /api/v1/catalog/providers returns provider metadata that can be used as the type of a future Linked Service. It does not return workspace credentials or configuration state.

GET /api/v1/catalog/providers

Property	JSON type	Description
`id`	String	Immutable provider identifier, such as `openai`.
`name`	String	Human-readable provider name.
`description`	String	Static provider description.
`logo_url`	String	Provider logo location.
`requires_api_key`	Boolean	Whether this provider type requires an API key.
`default_base_url`	String or `null`	Default endpoint for this provider type.
`model_catalog_note`	String or `null`	Static note about catalog support.
`offerings`	Array	Authoritative provider-owned model offerings.

Each offering includes model_id, provider_model_id, endpoint, optional context_window and max_output_tokens, optional pricing, status, and optional deprecated_at. Pricing includes input, output, cached-input, valid until, and note fields when known.

Example response (200 OK):

[
  {
    "id": "openai",
    "name": "OpenAI",
    "description": "OpenAI hosted models.",
    "logo_url": "/providers/openai.svg",
    "requires_api_key": true,
    "default_base_url": "https://api.openai.com/v1",
    "model_catalog_note": null,
    "offerings": [
      {
        "model_id": "gpt-5.6-sol",
        "provider_model_id": "gpt-5.6-sol",
        "endpoint": "responses",
        "context_window": 1050000,
        "max_output_tokens": 128000,
        "pricing": {
          "input_per_million_usd": "5.00",
          "output_per_million_usd": "30.00",
          "cached_input_per_million_usd": "0.50",
          "valid_until": null,
          "note": null
        },
        "status": "active",
        "deprecated_at": null
      }
    ]
  }
]

List model catalog

GET /api/v1/catalog/models returns active logical model metadata only. It does not identify provider routes, configured providers, or available workspace connections.

GET /api/v1/catalog/models

The response is an object with the following properties:

Property	JSON type	Description
`items`	Array	Active logical model definitions.
`total`	Integer	Total number of active logical model definitions.

Each items entry has the following properties:

Property	JSON type	Description
`id`	String	Immutable logical model identifier.
`name`	String	Human-readable model name.
`vendor`	String	Model vendor.
`family`	String	Model family.
`description`	String	Static model description.
`released_at`	String	ISO 8601 model release date.
`capabilities`	Array of strings	Catalog capability identifiers.
`status`	String	Static lifecycle status.
`tags`	Array of strings	Search and grouping tags.

Example response (200 OK):

{
  "items": [
    {
      "id": "gpt-5",
      "name": "GPT-5",
      "vendor": "OpenAI",
      "family": "GPT",
      "description": "A general-purpose model.",
      "released_at": "2026-08-07",
      "capabilities": [
        "coding",
        "text"
      ],
      "status": "active",
      "tags": [
        "general-purpose"
      ]
    }
  ],
  "total": 1
}

List built-in Persona catalog

GET /api/v1/catalog/personas returns packaged built-in Persona definitions. It does not return custom Personas, workspace enablement, or workspace action capabilities.

GET /api/v1/catalog/personas

Each array item includes its immutable id, name, description, complete instructions, applicability metadata, enabled_by_default, catalog_version, deprecated state, category, associated skill_ids, and the default_for agent roles.

Example response (200 OK):

[
  {
    "id": "builtin:persona:tech_lead",
    "name": "Tech Lead",
    "description": "Optimises for correct implementation and maintainable code.",
    "instructions": "# Persona: Tech Lead\n...",
    "metadata": {
      "applies_when": [
        "code",
        "implementation"
      ],
      "applies_to": [
        "generator",
        "reviewer"
      ]
    },
    "enabled_by_default": true,
    "catalog_version": 2,
    "deprecated": false,
    "category": "software_engineering",
    "skill_ids": [
      "builtin:skill:security"
    ],
    "default_for": [
      "generator"
    ]
  }
]

List built-in Skill catalog

GET /api/v1/catalog/skills returns packaged built-in Skill definitions. It does not return custom Skills, workspace enablement, or workspace action capabilities.

GET /api/v1/catalog/skills

Each array item includes the shared static instruction fields described for Personas, plus reviewer_default_strictness, phase, artifacts, and output when the Skill defines them.

Example response (200 OK):

[
  {
    "id": "builtin:skill:security",
    "name": "Security",
    "description": "Evaluates security risks.",
    "instructions": "# Security\n...",
    "metadata": {
      "applies_when": [
        "code",
        "implementation"
      ],
      "applies_to": [
        "reviewer"
      ]
    },
    "enabled_by_default": true,
    "catalog_version": 2,
    "deprecated": false,
    "category": "security_privacy",
    "reviewer_default_strictness": {
      "prototype": 40,
      "mvp": 65,
      "production": 85
    },
    "phase": "code_review",
    "artifacts": [
      "changed_files",
      "git_diff"
    ],
    "output": "json"
  }
]