Sub-Agent-Aware Routing

Sub-agent-aware routing keeps parent-agent traffic on one target while routing delegated sub-agent work across separate targets. Current support is available through the passthrough route's optional subagent_classifier table.

schema_version = 1

[llm_clients.openrouter]
format = "openai_chat"
base_url = "https://openrouter.ai/api/v1"
api_key_env = "OPENROUTER_API_KEY"

[targets.parent]
id = "anthropic/claude-sonnet-5"
llm_client = "openrouter"

[targets.classifier]
id = "openai/gpt-4.1-mini"
llm_client = "openrouter"

[targets.worker]
id = "openai/gpt-5.4-mini"
llm_client = "openrouter"

[targets.reviewer]
id = "anthropic/claude-opus-5"
llm_client = "openrouter"

[routes.agent]
id = "agent"
type = "passthrough"
target = "parent"
context_window = 400000
tool_calling = true
reasoning = true

[routes.agent.subagent_classifier]
classifier_target = "classifier"
targets = ["worker", "reviewer"]
default_target = "worker"
classify_trigger = "new_session"
max_output_tokens = 64
prompt = """
Select exactly one target for the delegated task.

- Select "reviewer" for code review, critique, auditing, or correctness analysis.
- Select "worker" for implementation, research, explanation, and other delegated work.

Return only JSON matching the response schema.
"""
response_schema = '''
{
  "type": "object",
  "properties": {
    "target": {"type": "string", "enum": ["worker", "reviewer"]}
  },
  "required": ["target"],
  "additionalProperties": false
}
'''
policy = { type = "target_selector", selector = "/target" }

Set OPENROUTER_API_KEY, save the configuration as routes.toml, and validate it before starting the server:

export OPENROUTER_API_KEY="sk-or-v1-..."  # pragma: allowlist secret
switchyard-server --config routes.toml --dry-run
switchyard-server --config routes.toml --port 4000

The parent always uses parent. For a delegated request, the classifier sees the prompt supplied by the parent and selects one configured target. With classify_trigger = "new_session", Switchyard reuses that decision for later requests from the same session + agent identity. Use every_request to classify each delegated request. user_turn is not supported for sub-agent routing. Harness-maintenance requests continue to the parent target.

Clients must still request the route ID (agent above). An explicit model name that is not registered as a route is rejected before sub-agent classification. message_hash_fallback is not supported for sub-agent routing because affinity requires harness-provided child identity.