CopilotKit

Agent Config

Forward typed configuration from your UI into the agent's reasoning loop.


"""PydanticAI agent backing the Agent Config Object demo.Reads three forwarded properties — tone, expertise, responseLength —from the AG-UI run's ``context`` (populated by the TS runtime route)and builds its system prompt dynamically per turn.PydanticAI-specific wiring--------------------------The CopilotKit provider's ``properties`` prop is forwarded by the runtimeas ``forwardedProps`` on each AG-UI run. PydanticAI's ``agent.to_ag_ui()``bridge surfaces that via ``ctx.deps.copilotkit.context`` when the runtimeroute repacks it (see ``src/app/api/copilotkit-agent-config/route.ts``)— the TS route appends a synthetic ``agent-config-properties`` contextentry whose JSON payload carries the three properties.A ``@agent.system_prompt`` dynamic prompt reads that context entry atcall-time and composes the system prompt from the three axes. When thecontext entry is missing (or contains an unknown value), we fall back tothe ``DEFAULT_*`` constants — same defensive behaviour as thelanggraph-python reference."""from __future__ import annotationsimport jsonfrom typing import Any, Literalfrom pydantic import BaseModelfrom pydantic_ai import Agent, RunContextfrom pydantic_ai.ag_ui import StateDepsfrom pydantic_ai.models.openai import OpenAIResponsesModelclass AgentConfigState(BaseModel):    """Agent-config demo carries no shared state — the provider properties    ride on ``context`` instead."""Tone = Literal["professional", "casual", "enthusiastic"]Expertise = Literal["beginner", "intermediate", "expert"]ResponseLength = Literal["concise", "detailed"]DEFAULT_TONE: Tone = "professional"DEFAULT_EXPERTISE: Expertise = "intermediate"DEFAULT_RESPONSE_LENGTH: ResponseLength = "concise"VALID_TONES: set[str] = {"professional", "casual", "enthusiastic"}VALID_EXPERTISE: set[str] = {"beginner", "intermediate", "expert"}VALID_RESPONSE_LENGTHS: set[str] = {"concise", "detailed"}PROPERTIES_CONTEXT_DESCRIPTION = "agent-config-properties"def _read_properties_from_context(    ctx: RunContext[StateDeps[AgentConfigState]],) -> dict[str, str]:    """Read the forwarded ``properties`` object with defensive defaults.    The TS runtime route at ``copilotkit-agent-config/route.ts`` appends a    context entry with ``description == "agent-config-properties"`` and a    JSON payload containing ``{tone, expertise, responseLength}``. Any    missing or unrecognized value falls back to the corresponding    ``DEFAULT_*`` constant. The function never raises.    """    copilotkit_state = getattr(ctx.deps, "copilotkit", None)    context_entries: list[Any] = []    if copilotkit_state and hasattr(copilotkit_state, "context"):        context_entries = copilotkit_state.context or []    payload: dict[str, Any] = {}    for entry in context_entries:        if not isinstance(entry, dict):            continue        if entry.get("description") != PROPERTIES_CONTEXT_DESCRIPTION:            continue        raw_value = entry.get("value")        if isinstance(raw_value, dict):            payload = raw_value        elif isinstance(raw_value, str):            try:                parsed = json.loads(raw_value)                if isinstance(parsed, dict):                    payload = parsed            except json.JSONDecodeError:                continue        if payload:            break    tone = payload.get("tone", DEFAULT_TONE)    expertise = payload.get("expertise", DEFAULT_EXPERTISE)    response_length = payload.get("responseLength", DEFAULT_RESPONSE_LENGTH)    if tone not in VALID_TONES:        tone = DEFAULT_TONE    if expertise not in VALID_EXPERTISE:        expertise = DEFAULT_EXPERTISE    if response_length not in VALID_RESPONSE_LENGTHS:        response_length = DEFAULT_RESPONSE_LENGTH    return {        "tone": tone,        "expertise": expertise,        "response_length": response_length,    }def _build_system_prompt(tone: str, expertise: str, response_length: str) -> str:    """Compose the system prompt from the three axes."""    tone_rules = {        "professional": ("Use neutral, precise language. No emoji. Short sentences."),        "casual": (            "Use friendly, conversational language. Contractions OK. "            "Light humor welcome."        ),        "enthusiastic": (            "Use upbeat, energetic language. Exclamation points OK. Emoji OK."        ),    }    expertise_rules = {        "beginner": "Assume no prior knowledge. Define jargon. Use analogies.",        "intermediate": (            "Assume common terms are understood; explain specialized terms."        ),        "expert": ("Assume technical fluency. Use precise terminology. Skip basics."),    }    length_rules = {        "concise": "Respond in 1-3 sentences.",        "detailed": ("Respond in multiple paragraphs with examples where relevant."),    }    return (        "You are a helpful assistant.\n\n"        f"Tone: {tone_rules[tone]}\n"        f"Expertise level: {expertise_rules[expertise]}\n"        f"Response length: {length_rules[response_length]}"    )agent = Agent(    model=OpenAIResponsesModel("gpt-4o-mini"),    deps_type=StateDeps[AgentConfigState],)@agent.system_promptdef build_prompt(ctx: RunContext[StateDeps[AgentConfigState]]) -> str:    props = _read_properties_from_context(ctx)    return _build_system_prompt(        props["tone"], props["expertise"], props["response_length"]    )__all__ = ["AgentConfigState", "agent"]

You have a working agent and want the user to be able to tune how it behaves: tone, expertise level, response length, language, persona. By the end of this guide, your UI will own a typed config object that the agent reads on every run and rebuilds its system prompt from.

When to use this#

Reach for agent config whenever the agent's behaviour depends on user-controllable settings that don't fit naturally as chat input:

  • Tone, voice, persona: "playful", "formal", "casual"
  • Expertise level: "beginner", "intermediate", "expert"
  • Response shape: short / medium / long, structured / prose, language
  • Domain switches: which knowledge base to consult, which tool subset to enable

If the values are a channel the user occasionally tunes (a settings panel, a toolbar of selects), agent config is the right shape. If the values are content the agent should write back to (notes, a document, a plan), use Shared State instead.

How agent config flows from the UI into the agent's reasoning loop depends on your runtime architecture. Agents living behind a runtime read it from agent state on every run, while in-process agents receive the same object as forwarded properties on the provider — same UX, slightly different wiring on each side.

How it works#

Agent config is a typed object the frontend owns and keeps in sync with the agent. There are two pieces: the UI side, which owns the React state and pushes every change into agent state, and the backend node, which reads those fields out of state and turns them into a system prompt.

The UI side stays simple. Hold the typed config in React state, then mirror every change into the agent through agent.setState({...}):

frontend/src/app/page.tsx — UI owns the typed config
function ConfigStateSync({ config }: { config: AgentConfig }) {
  const { agent } = useAgent({ agentId: "agent-config" });
  useEffect(() => {
    agent.setState({ ...config });
  }, [agent, config]);
  return null;
}

The backend half is also a single node. Read the config out of state at the top of every run and use it to build the system prompt for that turn:

backend/agent.py — agent reads config and rebuilds the system prompt
async def my_agent_node(state: AgentState, config: RunnableConfig):
    cfg = state.get("config", {})
    tone = cfg.get("tone", "casual")
    expertise = cfg.get("expertise", "intermediate")
    response_length = cfg.get("response_length", "medium")
    system_prompt = build_system_prompt(tone, expertise, response_length)
    # ...

The agent reads the latest typed config at the start of every turn, rebuilds the system prompt, runs the turn. This is the same shape as the shared-state write-side pattern; agent config is just a specific use of that pattern with a UI-owned typed object on top.