Pipecat

NiadraMemoryProcessor is a FrameProcessor that sits between aggregators.user() and the LLM. On every LLMContextFrame it reads the context and places it; observe(aggregators) records the turns. Python only: the Pipecat pipeline, where the model call happens, runs in Python; the JavaScript packages of Pipecat and Daily are browser clients, where a Niadra key must never go.

Install

pip install 'niadra[pipecat]'   # pipecat-ai 1.11.0 or newer, below 2; Python 3.11 or newer

The five primitives

Primitive	How the adapter wires it
Context	On each `LLMContextFrame`, between the user aggregator and the LLM, `context()` within 150 ms (the `voice` view); the pack goes right after the leading `system` or `developer` messages and the `turn_block` at the end. The blocks placed on the previous turn are taken out first, so the shared context never piles them up. A speculative inference gets them too, in its provisional copy	The read sends the user’s last turn as `query`; `memory.prefetcher()`, right after the STT service, sends the turn so far with `prefetch()` on each `InterimTranscriptionFrame` and `TranscriptionFrame`, holding no frame. The prefetch is reused only for the identical `query`; a longer transcript finds the memory already open, and a turn with time words or a count is recomputed at the read
Turns	`observe()` subscribes to the aggregators: each user message written to the context (`on_user_turn_message_added`, final in cascade and realtime modes) is the customer’s turn and each finished assistant turn (`on_assistant_turn_stopped`) is the agent’s. `EndFrame` and `CancelFrame` end the conversation
Tools	`history_tools()` gives the three history tools as `FunctionSchema`s that carry their own handlers, so the LLM service registers them from the context. Their JSON Schemas are the kit’s
Verification	`attestation=` with the carrier’s STIR/SHAKEN level (`A`, `B`, `C`, or Twilio’s `StirVerstat`), recorded once before the first context
Handoff	`transferred_to_human()` and `transferred_to_agent()` record the transfer; call them where the pipeline (or Pipecat Flows) hands the call over

conversation_for_call(niadra, call_sid, caller) opens the conversation with the CallSid as id and the customer’s number as subject.

Minimal example

"""A Pipecat phone agent over Twilio Media Streams with the customer's memory."""

from pipecat.pipeline.pipeline import Pipeline
from pipecat.processors.aggregators.llm_context import LLMContext
from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair

from niadra import AsyncNiadra
from niadra.integrations.pipecat import NiadraMemoryProcessor, conversation_for_call, history_tools

niadra = AsyncNiadra(channel="voice")


def build(transport, stt, llm, tts, call_sid: str, caller: str, stir_verstat: str | None) -> Pipeline:
    conversation = conversation_for_call(niadra, call_sid, caller)
    memory = NiadraMemoryProcessor(conversation, attestation=stir_verstat)
    context = LLMContext(
        [{"role": "system", "content": "You are Acme's agent."}], tools=history_tools(conversation)
    )
    aggregators = LLMContextAggregatorPair(context)
    memory.observe(aggregators)
    user, assistant = aggregators.user(), aggregators.assistant()
    return Pipeline([transport.input(), stt, user, memory, llm, tts, transport.output(), assistant])

The same code is in examples/pipecat_bot.py.

Agent memory

With agent_memory=True (or {"write": True, "max_tokens": 300, "tags": [...]}), the agent’s own notes go right before the customer’s context, in the same message, and history_tools(conversation, agent_memory=...) adds search_agent_memory (and remember). See Agent memory.

Limits

The processor always passes the frame on, even when Niadra fails: the pipeline never stops because of the memory.
It uses LLMContext and LLMContextFrame, Pipecat’s universal context since 1.9; OpenAILLMContext and LLMMessagesFrame are no longer in Pipecat’s main and are not handled.
Python 3.11 or newer only, because Pipecat asks for it. The pipecat and crewai extras pin incompatible versions of a shared dependency; install one per environment.
The shape of the processor follows Pipecat’s own Mem0 memory service (BSD 2-Clause, credited in the file); the reading is Niadra’s: one pinned pack per conversation, not a search per message.
Tested against pipecat-ai 1.11.0 with a real pipeline, a fake LLM and frames pushed by hand, without audio and without network.

Next steps

Voice agents

context before hello, network attestation and handoff.

Twilio

the call webhook and StirVerstat as proof.

Get started

Voice

Channels

Agent frameworks

Model SDKs

Automation platforms

MCP clients

Install

The five primitives

Minimal example

Agent memory

Limits

Next steps

Voice agents

Twilio

​Install

​The five primitives

​Minimal example

​Agent memory

​Limits

​Next steps

Voice agents

Twilio

Install

The five primitives

Minimal example

Agent memory

Limits

Next steps