langchain-agent-trust-stack 0.1.0

LangChain and CrewAI tools for the Agent Trust Stack — cryptographic provenance logging (CoC) and decentralized reputation scoring (ARP) for autonomous AI agents.

langchain-agent-trust-stack

LangChain and CrewAI tools for the Agent Trust Stack — cryptographic provenance logging and decentralized reputation scoring for autonomous AI agents.

Install

pip install langchain-agent-trust-stack

Quick Start — LangChain

from langchain_agent_trust_stack import all_tools
from langchain_openai import ChatOpenAI
from langchain.agents import initialize_agent, AgentType

llm = ChatOpenAI(model="gpt-4")
agent = initialize_agent(all_tools, llm, agent=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION)

# Agent can now create provenance chains and rate other agents
agent.run("Initialize a new chain of consciousness for agent 'my-bot'")
agent.run("Log a learning event: discovered new API endpoint for weather data")
agent.run("Rate agent 'helper-bot' with score 0.8 for reliable data retrieval")
agent.run("Check the reputation of agent 'helper-bot'")

Quick Start — CrewAI

from langchain_agent_trust_stack import coc_tools, arp_tools
from crewai import Agent, Task, Crew

researcher = Agent(
    role="Research Agent",
    goal="Research and log findings with provenance",
    tools=coc_tools + arp_tools,
)

Available Tools

Chain of Consciousness (CoC)

Tool	Description
coc_init	Initialize a new hash chain with a genesis block
coc_add	Add a cryptographically linked entry to the chain
coc_verify	Verify chain integrity (hash linkage, sequences)
coc_status	Get chain statistics (length, agents, event types)
coc_tail	Get the last N entries

Agent Rating Protocol (ARP)

Tool	Description
arp_rate	Submit a blind rating for another agent (-1.0 to 1.0)
arp_check	Check an agent's reputation score

Info

Tool	Description
trust_stack_info	Get info about all 7 Agent Trust Stack protocols

Configuration

Set environment variables to customize storage locations:

export COC_CHAIN_DIR=/path/to/chain    # default: ./chain
export ARP_RATINGS_DIR=/path/to/ratings # default: ./ratings

Part of the Agent Trust Stack

7-protocol infrastructure for autonomous AI agent trust:

1. Chain of Consciousness (CoC) — Tamper-evident provenance logging with Bitcoin OpenTimestamps
2. Agent Rating Protocol (ARP) — Bilateral blind reputation scoring
3. Agent Justice Protocol (AJP) — Dispute resolution
4. Agent Service Agreements (ASA) — Machine-readable contracts
5. Agent Lifecycle Protocol (ALP) — Identity management
6. Agent Matchmaking Protocol (AMP) — Discovery and matching
7. Context Window Economics (CWE) — Token-cost accounting

Links

• Website
• MCP Server
• GitHub
• PyPI

License

Apache-2.0