Update complexity analyzer config

Replaces the full complexity analyzer runtime config and hot-reloads the routing plugin. Changing the embedding configuration or reference phrases triggers a background re-warm of the semantic classifier; unchanged phrases are not re-embedded. Setting semantic.fallback to llm requires an llm block. A decision model can be selected as the primary classifier or semantic fallback; it uses the decision block’s provider and model (default typesafe/jev-latest) and that provider’s credentials.

PUT /api/routing/complexity-analyzer-config

Error

A valid request URL is required to generate request examples

200

{
  "keywords": {
    "simple_keywords": [
      "<string>"
    ],
    "medium_keywords": [
      "<string>"
    ],
    "complex_keywords": [
      "<string>"
    ]
  },
  "tier_boundaries": {
    "simple_medium": 123,
    "medium_complex": 123
  },
  "classifier": "semantic",
  "semantic": {
    "provider": "<string>",
    "embedding_model": "<string>",
    "timeout": "<string>",
    "min_similarity": 1,
    "message_history_count": 5,
    "count_toward_budgets": true,
    "vector_store": "embedded",
    "fallback": "none"
  },
  "decision": {
    "provider": "<string>",
    "model": "<string>",
    "previous_message_count": 1,
    "timeout": "<string>",
    "criteria": {
      "SIMPLE": {
        "definition": "<string>",
        "signals": [
          "<string>"
        ],
        "examples": [
          "<string>"
        ]
      },
      "MEDIUM": {
        "definition": "<string>",
        "signals": [
          "<string>"
        ],
        "examples": [
          "<string>"
        ]
      },
      "COMPLEX": {
        "definition": "<string>",
        "signals": [
          "<string>"
        ],
        "examples": [
          "<string>"
        ]
      }
    }
  },
  "llm": {
    "provider": "<string>",
    "model": "<string>",
    "timeout": "<string>",
    "prompt": "<string>",
    "message_history_count": 5,
    "count_toward_budgets": true
  },
  "session": {
    "enabled": true
  }
}

Required Permissions

Permission	Granted via
Governance:Update	RBAC role or API key scope

How permissions work

Authorizations

ManagementBearerAuthBasicAuth

Authorization · string · header · required

Management API authentication for /api/* endpoints. Use the Authorization header with Bearer <token>, where <token> is one of:

• a Bifrost management API key,
• a dashboard session token issued by POST /api/session/login,
• base64 of <admin-username>:<admin-password> (legacy equivalent of BasicAuth).

Virtual keys (sk-bf-*) and the x-api-key header are not accepted on management APIs - the sole exception is GET /api/governance/virtual-keys/quota, which is virtual-key-only.

Authentication alone is not sufficient in Bifrost Enterprise: each operation page shows a Required Permissions table (Resource:Operation, for example Dashboard:View) above its Authorizations section, and the caller's RBAC role or management API key scopes must include what it lists, otherwise the request is rejected with 403 Forbidden.

A local admin — authenticated with the admin password, or any caller on a deployment with dashboard auth disabled — bypasses these checks and can call every management endpoint.

OSS setup lock. On Bifrost OSS, while dashboard auth is not active (no admin account, or auth disabled), every management endpoint except the public ones (/health, /api/version, /api/session/is-auth-enabled, /api/session/login, ...) requires the operator's setup token in the X-Bifrost-Setup-Token header, in place of Authorization. The token is set with setup_token in config.json or the BIFROST_SETUP_TOKEN environment variable. A missing header returns 401, a wrong token 403. The header stops working once dashboard auth is enabled. The dashboard instead trades the token once for an HttpOnly bifrost_setup_session cookie via POST /api/session/setup. See Required permissions for how permissions are derived and which endpoints are exempt.

Body

application/json

Full runtime configuration for complexity routing analysis.

keywords · object · required

Reference phrases for semantic complexity classification, one list per tier. Each request is embedded and takes the tier of its nearest phrase. Entries should be whole example prompts rather than individual keywords. Each phrase is limited to 2000 characters, the normalized lists may contain at most 750 phrases combined, and the same phrase cannot appear in two tiers. The field names predate semantic classification and are kept for backward compatibility.

Show child attributes

tier_boundaries · object · deprecated

Deprecated score thresholds for the retired lexical complexity classifier. This block is optional and ignored by semantic classification, but remains accepted for backward compatibility. When supplied, values must satisfy 0 < simple_medium < medium_complex < 1.

Show child attributes

classifier · enum<string> · default: semantic

Primary complexity classifier; omitted configuration keeps the semantic default.

Available options: semantic, decision

semantic · object

Embedding configuration for semantic complexity classification. When absent, no complexity tier is published and rules referencing complexity_tier fall through. The provider must have an enabled key.

Show child attributes

decision · object

Decision-model complexity-classifier request settings. Credentials come from the selected provider's configuration.

Show child attributes

llm · object

Chat-completion (LLM) fallback classifier settings. The block is inert unless semantic.fallback selects llm; the classifier runs only after semantic classification produces no tier. Tier names (SIMPLE, MEDIUM, COMPLEX) and the response format are fixed: the gateway always appends a non-editable reinforcement stating them.

Show child attributes

session · object

Session-aware complexity routing. When enabled and a supported session identity is present, Bifrost retains the highest tier observed across normally sequential turns for 24 hours of inactivity. It uses the configured primary classifier and is independent of provider session stickiness.

Show child attributes

Response

200

application/json

Complexity analyzer config updated successfully

Full runtime configuration for complexity routing analysis.

keywords · object · required

Reference phrases for semantic complexity classification, one list per tier. Each request is embedded and takes the tier of its nearest phrase. Entries should be whole example prompts rather than individual keywords. Each phrase is limited to 2000 characters, the normalized lists may contain at most 750 phrases combined, and the same phrase cannot appear in two tiers. The field names predate semantic classification and are kept for backward compatibility.

Show child attributes

tier_boundaries · object · deprecated

Deprecated score thresholds for the retired lexical complexity classifier. This block is optional and ignored by semantic classification, but remains accepted for backward compatibility. When supplied, values must satisfy 0 < simple_medium < medium_complex < 1.

Show child attributes

classifier · enum<string> · default: semantic

Primary complexity classifier; omitted configuration keeps the semantic default.

Available options: semantic, decision

semantic · object

Embedding configuration for semantic complexity classification. When absent, no complexity tier is published and rules referencing complexity_tier fall through. The provider must have an enabled key.

Show child attributes

decision · object

Decision-model complexity-classifier request settings. Credentials come from the selected provider's configuration.

Show child attributes

llm · object

Chat-completion (LLM) fallback classifier settings. The block is inert unless semantic.fallback selects llm; the classifier runs only after semantic classification produces no tier. Tier names (SIMPLE, MEDIUM, COMPLEX) and the response format are fixed: the gateway always appends a non-editable reinforcement stating them.

Show child attributes

session · object

Session-aware complexity routing. When enabled and a supported session identity is present, Bifrost retains the highest tier observed across normally sequential turns for 24 hours of inactivity. It uses the configured primary classifier and is independent of provider session stickiness.

Show child attributes