Existing agent frameworks can keep their orchestration layer and gain a stronger communication layer.
`bca2p` is designed to sit above tools, agents, graph runtimes, and transport protocols as a semantic coordination layer: typed receptors, scoped signaling, complexes, quorum, homeostasis, and causal feedback.
| Layer | What bca2p adds | Best fit |
|---|---|---|
| Semantic layer | Signals, receptors, scopes, complexes, quorum, and homeostasis. | All frameworks |
| Runtime layer | Native execution, graph adapters, checkpointing, and replay. | LangGraph, native bca2p |
| Tool and agent layer | Subagent wrappers, tool metadata, typed handoffs, and middleware. | LangChain, CrewAI, Google ADK |
| Transport bridge | A2A-compatible message exchange with signal metadata and artifacts. | A2A Protocol, remote agents |
| Learning layer | Causal feedback, counterfactual replay, and communication policy updates. | Any traceable runtime |
Agent API and tool-first composition
Use `bca2p` to replace implicit subagent messaging with typed signals, receptor-aware tool invocation, quorum-driven escalation, and causal metadata on each handoff.
structured subagent tool contracts
signal-aware middleware
causal feedback on tool and agent routing
Stateful graph orchestration and replay
Use `bca2p` to bring biological communication semantics into graph nodes, channels, checkpointing, and replay rather than treating communication as ordinary state mutation.
receptor-bound routing in graph nodes
complex formation over graph branches
signal lineage layered onto existing replay
Role-oriented multi-agent teams
Use `bca2p` to give role-based crews explicit signaling modes, scoped dissemination, and homeostatic damping instead of relying on role prompts alone.
clear local vs global crew signaling
quorum-based coordination semantics
causal diagnosis of crew behavior
Interoperable agent-to-agent transport
Use `bca2p` as the semantic layer above transport so independent agents exchange typed biological signals, artifacts, and causal context rather than opaque payloads.
transport plus signaling semantics
artifact-aware vesicle-style exchange
cross-system signal normalization
Modular agent development and orchestration
Use `bca2p` to define receptor contracts, topology policy, quorum, and causal feedback around ADK agents, tools, sessions, and workflow agents.
typed communication around ADK agents
state-aware signaling policies
research-grade route analysis
The biological layer is not decorative. It gives framework users concrete coordination primitives.
Cell signaling as a software design primitive
Autocrine, paracrine, endocrine, juxtacrine, and synaptic patterns map cleanly to self-regulation, local coordination, global broadcast, direct contact, and persistent weighted links between agents.
Causal inference as a coordination primitive
Signals are not just logged; they can be attributed, replayed, damped, amplified, and compared counterfactually so the communication policy evolves with evidence.
System homeostasis as an engineering primitive
The SDK treats overload, route instability, retry storms, and noisy senders as first-class phenomena that the system should regulate automatically.
middleware = BioAgentMiddleware(agent_id="planner")
signal = middleware.build_signal_for_tool(
agent_id="billing_agent",
recipient_scope="billing",
payload={"ticket_id": "T-44", "priority": "high"},
)
# receptor-aware routing
# quorum-based escalation
# causal metadata on each handoffgraph = BioGraph()
graph.add_agent("planner", planner_handler)
graph.add_quorum_rule(...)
adapter = LangGraphAdapter(graph)
compiled = adapter.to_state_graph()
# replay remains useful
# communication gains biological semanticsrequest = A2ABridge.signal_to_send_request(signal)
# A2A moves the payload
# bca2p defines the signal meaning
# artifact and causal metadata survive the hop