boltzstack-engine 0.0.1

Reusable BoltzStack agent engine runtime.

BoltzStack Engine

boltzstack-engine is the reusable agent runtime for BoltzStack. It provides the model loop, tool registry, context planning, hydration, artifact handles, scoped memory, run traces, provider adapters, and sandbox runtime contracts.

Development

Install the package in editable mode with development dependencies:

make install

Run the smoke tests:

make test

Run the full local check suite:

make check

Minimal Usage

from boltzstack_engine import AgentDefinition, AgentState
from boltzstack_engine.model_adapters.openai import openai_model_client_from_env

agent = AgentDefinition(
    name="support_agent",
    system_prompt="Answer with concise, evidence-backed responses.",
)
model_client = openai_model_client_from_env(system_prompt=agent.system_prompt)

if model_client is None:
    raise RuntimeError("No model provider is configured")

loop = agent.create_loop(model_client)
output = await loop.run("Summarize the latest run.", AgentState(), context={})
print(output.text)

Use the boltzstack_engine package namespace and BoltzStack Engine terminology in docs and integrations.