agentproofx 0.3.0

Static graph verification for agent workflows — LangGraph, ADK, AutoGen, CrewAI

Agentproof

Static graph verification for agent workflows. Prove safety properties on your workflow graph before deployment — no runtime overhead, no gatekeeping layer.

Supports LangGraph, Google ADK, AutoGen, and CrewAI.

Install

pip install agentproof

With framework extractors:

pip install agentproof[langgraph]
pip install agentproof[adk]
pip install agentproof[autogen]
pip install agentproof[crewai]
pip install agentproof[all-frameworks]

Quick Start

from agentproof import verify
from agentproof.graph.model import (
    AgentGraph, GraphNode, GraphEdge, NodeKind, EdgeKind,
)

# Build a graph (or extract one from a framework — see below)
graph = AgentGraph(
    name="my_pipeline",
    framework="manual",
    nodes=(
        GraphNode(id="entry", kind=NodeKind.ENTRY, label="start"),
        GraphNode(id="fetch", kind=NodeKind.TOOL, label="Fetch", tools=("http_get",)),
        GraphNode(id="review", kind=NodeKind.HUMAN, label="Human Review"),
        GraphNode(id="store", kind=NodeKind.TOOL, label="Store", tools=("db_insert",)),
        GraphNode(id="exit", kind=NodeKind.EXIT, label="end"),
    ),
    edges=(
        GraphEdge(source="entry", target="fetch"),
        GraphEdge(source="fetch", target="review"),
        GraphEdge(source="review", target="store"),
        GraphEdge(source="store", target="exit"),
    ),
    entry_id="entry",
    exit_ids=("exit",),
)

report = verify(graph, require_human=True)
print(report["structural"])

Features

• Structural checks — reachability, dead-end detection, human-in-the-loop enforcement, tool declaration coverage, router edge validation, entry/exit structure
• Temporal verification (LTL) — define safety rules in a lightweight LTL DSL (e.g. G !tool:rm_rf), compile to DFA monitors, and verify against simulated traces or statically via graph-DFA product construction
• Framework extractors — convert native workflow objects from LangGraph, ADK, AutoGen, and CrewAI into Agentproof's framework-agnostic AgentGraph
• Trace generation — random-walk trace generator for temporal policy testing
• Typed — full type annotations, py.typed marker

Supported Frameworks

Framework	Extra	Extractor
LangGraph	agentproof[langgraph]	agentproof.graph.extract._langgraph
Google ADK	agentproof[adk]	agentproof.graph.extract._adk
AutoGen	agentproof[autogen]	agentproof.graph.extract._autogen
CrewAI	agentproof[crewai]	agentproof.graph.extract._crewai

You can also construct an AgentGraph directly without any framework dependency.

Examples

See the examples/ directory for complete walkthroughs:

• full_verification.py — end-to-end structural + temporal verification (no framework needed)
• langgraph_customer_support.py — extract and verify a LangGraph agent
• adk_pipeline.py — Google ADK extraction
• autogen_debate.py — AutoGen team verification
• crewai_research_crew.py — CrewAI crew verification

License

MIT