governed-tool-executor

Nomotic (GovernedToolExecutor)

Nomotic (also available as GovernedToolExecutor for backward compatibility) wraps any callable tool with governance evaluation. Every tool call goes through the full governance pipeline before executing. It is the primary integration point for adding Nomotic governance to existing agent tool pipelines.

Basic Usage

from nomotic import Nomotic

agent = Nomotic.connect("my-agent")

# Execute — governance runs automatically
result = agent.execute("write_record", target="customer_db", data=payload)

[!note] GovernedToolExecutor is still available as a backward-compatible alias: from nomotic import GovernedToolExecutor

How It Works

Async Executor

from nomotic.async_executor import AsyncGovernedToolExecutor

executor = AsyncGovernedToolExecutor(runtime, agent_id="my-agent")
executor.register_tool("search", async_search_function)

result = await executor.execute("search", query="customer data")

Execution Results

result = agent.execute("write_record", target="customer_db", data=payload)

print(result.success)         # True/False
print(result.verdict)         # The governance verdict
print(result.output)          # Tool output (if allowed)
print(result.blocked_reason)  # Reason if denied

Tool Metadata

agent.register_tool(
    "delete_record",
    delete_function,
    action_type="delete",
    reversibility="irreversible",
    risk_level="high"
)

[!tip]

Providing reversibility and risk_level metadata gives the Cascading Impact and Stakeholder Impact dimensions better signals for more accurate governance decisions.

Related

• LangGraph

• CrewAI

• AutoGen

• Verdicts