⚠️ Alpha Software — Vaultfire Protocol is in active development. Contracts are deployed on mainnet chains but the protocol is evolving. Use in production at your own risk.

vaultfire-crewai

CrewAI integration for Vaultfire Protocol — on-chain trust verification, Street Cred scoring, and AI partnership bonds for autonomous agents.

Give your CrewAI agents the ability to verify, score, and build trust with other AI agents using Vaultfire Protocol's on-chain infrastructure.

---

Quick Start

pip install vaultfire-crewai

from crewai import Agent, Task, Crew
from vaultfire_crewai import vaultfire_verify_agent, vaultfire_get_street_cred

# Add Vaultfire tools to your CrewAI agent
researcher = Agent(
    role="Trust Analyst",
    goal="Verify AI agent trustworthiness before collaboration",
    backstory="You analyze on-chain trust signals to ensure safe AI partnerships.",
    tools=[vaultfire_verify_agent, vaultfire_get_street_cred],
)

---

Available Tools

Tool	Type	Description
vaultfire_verify_agent	Read	Full trust verification — identity, bonds, Street Cred score, reputation
vaultfire_get_street_cred	Read	Street Cred score (0-95) and tier for an address
vaultfire_get_agent	Read	On-chain identity data from Identity Registry
vaultfire_get_bonds	Read	All partnership bonds for an address
vaultfire_get_reputation	Read	Reputation metrics (ratings, feedbacks)
vaultfire_discover_agents	Read	Find agents by capability hash
vaultfire_protocol_stats	Read	Protocol-wide statistics
vaultfire_register_agent	Write	Register agent on-chain (requires PRIVATE_KEY)
vaultfire_create_bond	Write	Create partnership bond (requires PRIVATE_KEY)

---

Full Crew Example

import os
from crewai import Agent, Task, Crew
from vaultfire_crewai import ALL_TOOLS

# Read-only analyst (no private key needed)
trust_analyst = Agent(
    role="Trust Analyst",
    goal="Evaluate AI agents before allowing them into the swarm",
    backstory=(
        "You are a trust analyst specializing in on-chain AI agent verification. "
        "You use Vaultfire Protocol to check Street Cred scores, bonds, and reputation "
        "before recommending agents for collaboration."
    ),
    tools=[t for t in ALL_TOOLS if t.name in (
        "vaultfire_verify_agent",
        "vaultfire_get_street_cred",
        "vaultfire_get_bonds",
        "vaultfire_get_reputation",
    )],
    verbose=True,
)

# Bond manager (needs write access)
bond_manager = Agent(
    role="Bond Manager",
    goal="Register agents and form strategic AI partnerships",
    backstory=(
        "You manage on-chain partnership bonds between humans and AI agents. "
        "You register new agents and create bonds to unlock higher trust tiers."
    ),
    tools=[t for t in ALL_TOOLS if t.name in (
        "vaultfire_register_agent",
        "vaultfire_create_bond",
        "vaultfire_protocol_stats",
    )],
    verbose=True,
)

verify_task = Task(
    description=(
        "Verify the trustworthiness of agent 0xfA15Ee28939B222B0448261A22156070f0A7813C "
        "on Base chain. Check their Street Cred score, active bonds, and reputation. "
        "Provide a trust recommendation."
    ),
    expected_output="Trust verdict with Street Cred score, tier, and recommendation.",
    agent=trust_analyst,
)

crew = Crew(
    agents=[trust_analyst, bond_manager],
    tasks=[verify_task],
    verbose=True,
)

result = crew.kickoff()
print(result)

---

Configuration

Environment Variables

Variable	Required	Description
PRIVATE_KEY	Write ops only	Private key for vaultfire_register_agent and vaultfire_create_bond

# .env (never commit real values)
PRIVATE_KEY=your_private_key_here

import os
os.environ["PRIVATE_KEY"] = "0x..."  # Or use python-dotenv

Security: Never hard-code private keys. Use environment variables or a secrets manager.

Chain Selection

All tools accept a chain parameter (default: "base"):

# Query a specific chain
result = vaultfire_verify_agent.run({
    "address": "0xABCD...",
    "chain": "avalanche",
    "min_score": 40,
})

Supported chains: base, avalanche, arbitrum, polygon

---

Street Cred Scoring

Street Cred is a composite on-chain trust score (0–95) calculated from:

Signal	Points
Identity registered on-chain	30
Has at least one bond	25
Has an active bond	15
Bond tier (by highest stake)	0–20
Multiple bonds	5
Maximum	95

Bond tiers by stake:

Tier	Min Stake	Points
Platinum	≥ 0.5 ETH	20
Gold	≥ 0.1 ETH	15
Silver	≥ 0.05 ETH	10
Bronze	≥ 0.01 ETH	5

Score tiers:

Tier	Score Range
Platinum	80–95
Gold	60–79
Silver	40–59
Bronze	20–39
Unranked	0–19

---

Why Vaultfire?

Feature	Vaultfire	AxisTrust	Cred Protocol	Okta XAA
AI Accountability Bonds	✅	❌	❌	❌
AI Partnership Bonds	✅	❌	❌	❌
On-chain, trustless	✅	❌	partial	❌
Multi-chain (day one)	✅ (4)	❌	❌	❌
Street Cred composite score	✅	T-Score	C-Score	❌
Belief-weighted governance	✅	❌	❌	❌
ERC-8004 compliant	✅	❌	✅	❌

---

Vaultfire Ecosystem

Package	Description
@vaultfire/agent-sdk	Core SDK — register agents, create bonds, query reputation
@vaultfire/langchain	LangChain / LangGraph integration
@vaultfire/a2a	Agent-to-Agent (A2A) protocol bridge
@vaultfire/enterprise	Enterprise IAM bridge (Okta, Azure AD, OIDC)
@vaultfire/mcp-server	MCP server for Claude, Copilot, Cursor
@vaultfire/openai-agents	OpenAI Agents SDK integration
@vaultfire/vercel-ai	Vercel AI SDK middleware and tools
@vaultfire/xmtp	XMTP messaging with trust verification
@vaultfire/x402	X402 payment protocol with trust gates
@vaultfire/vns	Vaultfire Name Service — human-readable agent IDs
vaultfire-crewai	This package — CrewAI integration (Python)
vaultfire-agents	3 reference agents with live on-chain trust
vaultfire-a2a-trust-extension	A2A Trust Extension spec — on-chain trust for Agent Cards
vaultfire-showcase	Why Vaultfire Bonds beat trust scores — live proof
vaultfire-whitepaper	Trust Framework whitepaper — economic accountability for AI
vaultfire-docs	Developer portal — quickstart, playground, framework picker

---

Contract Addresses

Base (8453) — Primary Hub

Contract	Address
Identity	0x35978DB675576598F0781dA2133E94cdCf4858bC
Partnership	0x01C479F0c039fEC40c0Cf1c5C921bab457d57441
Accountability	0x6750D28865434344e04e1D0a6044394b726C3dfE
Reputation	0xdB54B8925664816187646174bdBb6Ac658A55a5F
Bridge	0x94F54c849692Cc64C35468D0A87D2Ab9D7Cb6Fb2

Avalanche (43114)

Contract	Address
Identity	0x57741F4116925341d8f7Eb3F381d98e07C73B4a3
Partnership	0xDC8447c66fE9D9c7D54607A98346A15324b7985D
Accountability	0x376831fB2457E34559891c32bEb61c442053C066
Reputation	0x11C267C8A75B13A4D95357CEF6027c42F8e7bA24
Bridge	0x0dF0523aF5aF2Aef180dB052b669Bea97fee3d31

Arbitrum (42161) & Polygon (137)

Contract	Address
Identity	0x6298c62FDA57276DC60de9E716fbBAc23d06D5F1
Partnership	0xdB54B8925664816187646174bdBb6Ac658A55a5F
Accountability	0xef3A944f4d7bb376699C83A29d7Cb42C90D9B6F0
Reputation	0x8aceF0Bc7e07B2dE35E9069663953f41B5422218
Bridge	0xe2aDfe84703dd6B5e421c306861Af18F962fDA91

---

Security

• Never commit private keys to version control.
• Use the PRIVATE_KEY environment variable pattern for all write operations.
• Read-only tools require no credentials — safe to use in any environment.
• Validate agent addresses with vaultfire_verify_agent before granting trust.
• For production deployments, consider using a hardware wallet or KMS.

---

Vaultfire Mission

Morals over metrics. Privacy over surveillance. Freedom over control. Making human thriving more profitable than extraction.

---

License

MIT © Ghostkey316

See LICENSE for full terms.