Specialised Agents¶

Four agents for use cases beyond standard DeFi - DAO governance integrity, GameFi player verification, trust scoring for autonomous AI agent wallets, and ERC-8004 registry-based agent trust screening.

chainaware-governance-screener¶

Model: claude-haiku-4-5-20251001 | Tools: predictive_behaviour, predictive_fraud, predictive_behaviour_batch, predictive_fraud_batch, check_job_status, get_job_results

DAO voter screening - Sybil detection, governance tier classification, and voting weight multiplier calculation. Supports token-weighted, reputation-weighted, and quadratic governance models.

Prevents governance attacks where coordinated low-quality wallets accumulate voting power through token concentration without genuine community participation.

Example prompts¶

Should this wallet be allowed to vote? 0xVoter... on Ethereum

Voting weight for 0xMember... on BNB Chain - we use reputation-weighted governance

Screen our DAO members before the next vote: [list of addresses] on Ethereum

Sybil check for governance - are any of these addresses controlled by the same actor?
[list] on Ethereum

Filter fake voters from our proposal - here are all wallets that voted: [addresses]

Governance health check for our DAO: [member addresses] on Base
Governance model: quadratic

Governance tiers and multipliers¶

Tier	Multiplier	Criteria
Core Contributor	2×	Veteran, high experience, clean AML
Active Member	1.5×	Intermediate+, active participation
Participant	1×	Basic eligibility, legitimate wallet
Observer	0.5×	Low experience, below threshold
Disqualified	0×	Fraud flags, Sybil detection, bots

Example output¶

GOVERNANCE SCREENING
Wallet: 0xVoter... | Network: Ethereum
Governance Model: Reputation-weighted

Tier: ✅ Core Contributor
Voting Weight Multiplier: 2×
Sybil Risk: None detected

Profile:
  Experience:      Veteran (3.6 years on-chain)
  Fraud risk:      Very Low (0.03)
  AML status:      Clean
  On-chain governance history: 12 prior votes across 4 DAOs

Recommendation: Full voting rights. Eligible for governance committee
                nomination based on participation history.

GOVERNANCE SCREENING
Wallet: 0xSuspect... | Network: Ethereum

Tier: 🚫 DISQUALIFIED
Voting Weight Multiplier: 0×
Sybil Risk: HIGH

Flags:
  - Wallet created 8 days ago ⚠️
  - 3 similar wallets with near-identical creation patterns detected ⚠️
  - Token balance acquired in single transaction (typical Sybil pattern) ⚠️
  - No prior governance participation

Recommendation: Block from voting. High probability of Sybil attack.
                Flag the 3 related addresses for review.

GOVERNANCE HEALTH CHECK
DAO Members: 200 wallets | Network: Ethereum

DISTRIBUTION
  Core Contributors:  28 (14%)  - 2× weight
  Active Members:     61 (31%)  - 1.5× weight
  Participants:       74 (37%)  - 1× weight
  Observers:          22 (11%)  - 0.5× weight
  Disqualified:       15 (8%)   - 0× weight

Governance Health Score: 72 / 100 - Good

⚠️ Notable findings:
  - 4 address clusters detected (possible coordinated Sybil attack)
  - 15% of voting weight concentrated in 3 wallets (centralisation flag)
  - Recommend: minimum 90-day wallet age for new membership applications

chainaware-gamefi-screener¶

Model: claude-haiku-4-5-20251001 | Tools: predictive_fraud, predictive_behaviour

Web3 game and Play-to-Earn wallet screening. Detects bot farms, multi-account cheaters, and reward abusers. Classifies legitimate players into experience tiers for matchmaking. Calculates P2E reward eligibility and multipliers.

Example prompts¶

Is this wallet a real player or a bot? 0xPlayer... on BNB Chain

P2E eligibility for 0xGamer... on Ethereum
Our game: RPG with $TOKEN rewards, max $500/month per player

Bot detection for my game - screen these 100 new player wallets: [list]

What matchmaking tier for this wallet? 0x... on Polygon

Is this a farm wallet? They're claiming rewards from 4 different accounts: [list]

Detect cheaters in my P2E platform - unusual reward patterns:
Suspicious wallets: [list] on BSC

Reward eligibility for these 200 tournament participants: [list]
Token cap: $1,000 per player | Network: Ethereum

Player verdicts¶

Verdict	Meaning
✅ ALLOW	Legitimate player, eligible for full rewards
⚠️ FLAG	Suspicious signals, reduce rewards or require verification
🚫 BLOCK	Bot, farm wallet, or known cheater

Experience tiers (for matchmaking)¶

Tier	Label	Profile
T1	Veteran Gamer	High experience, 2+ years on-chain
T2	Experienced	Intermediate, 6-24 months
T3	Casual	Beginner, <6 months
T4	New Player	Very limited history

Example output¶

GAMEFI SCREENING
Wallet: 0xPlayer... | Network: BNB Chain | Game: RPG

Player Verdict: ✅ ALLOW

Legitimacy Score: 87 / 100
Experience Tier: T2 - Experienced

P2E Reward Eligibility: ✅ Full rewards
Reward Multiplier: 1.0× (standard)
Matchmaking Bracket: Competitive (Tier 2 pool)

Profile:
  On-chain age: 14 months
  Behaviour: Active trader, some gaming history
  Fraud risk: Very Low (0.04)
  Bot indicators: None detected

GAMEFI SCREENING
Wallet: 0xFarm001... | Network: BNB Chain

Player Verdict: 🚫 BLOCK

Legitimacy Score: 12 / 100

Bot indicators:
  ⚠️ Wallet created 3 days ago
  ⚠️ Transaction pattern matches known bot signature (high frequency, small amounts)
  ⚠️ 7 related wallets with identical creation patterns detected
  ⚠️ No organic DeFi activity - only game interactions

Recommendation: Block all 7 related addresses.
                This is a coordinated farm operation.

TOURNAMENT REWARD ELIGIBILITY (200 participants)

RESULTS
  ✅ Full rewards eligible:    143 (72%)
  ⚠️ Reduced rewards (flagged): 29 (15%)
  🚫 Blocked (bots/farms):      28 (14%)

Bot farm detected: 12-wallet cluster (0xFarm001...-012...)
Recommend: Ban cluster + review IP/device data for additional accounts

LEGITIMATE REWARD DISTRIBUTION
  T1 Veteran:   18 players × $1,000 = $18,000
  T2 Experienced: 61 players × $600 = $36,600
  T3 Casual:    64 players × $250  = $16,000
  Total:                             $70,600 (of $200,000 cap)
  Savings from bot removal:          $129,400

chainaware-agent-screener¶

Model: claude-haiku-4-5-20251001 | Tools: predictive_fraud, predictive_behaviour, predictive_rug_pull

Screens an AI agent's operational wallet and feeder wallet for trustworthiness. As autonomous AI agents transact on-chain, verifying the wallet behind an agent becomes as important as verifying a human counterparty.

Agent Trust Score formula: 0 = confirmed fraud, 1 = new/insufficient data, 2-10 = normalised reputation score.

Example prompts¶

Is this agent wallet safe? 0xAgentWallet... on Ethereum

Screen this AI agent before I allow it to interact with my protocol:
Agent wallet: 0xAgent... Feeder wallet: 0xFeeder... on Base

Check the feeder wallet for this agent - who's funding it? 0xFeeder... on ETH

Can I trust this agent? It wants to execute trades on my behalf: 0xAgent...

Agent trust score for this operator wallet: 0x... on Ethereum

Screen these 5 AI agents before we allow them into our protocol: [list of agent+feeder pairs]

Trust score reference¶

Score	Meaning
0	Confirmed fraud - do not interact
1	Insufficient data (new wallet) - cannot assess
2-4	Low trust - proceed with caution
5-7	Moderate trust - standard interaction
8-10	High trust - clean history, experienced operator

Example output¶

AGENT SCREENING
Agent Wallet:  0xAgent...  | Network: Ethereum
Feeder Wallet: 0xFeeder... | Network: Ethereum

Agent Trust Score: 8.4 / 10 ✅

Agent Wallet:
  Fraud verdict:   Not Fraud (0.03)
  On-chain age:    11 months
  Behaviour:       Consistent, programmatic - expected agent pattern
  AML flags:       None

Feeder Wallet:
  Fraud verdict:   Not Fraud (0.05)
  Experience:      High (Veteran operator wallet)
  AML flags:       None

Overall Recommendation: ✅ Trust. Agent and feeder both show clean histories.
                         Operator appears to be a legitimate, experienced actor.

AGENT SCREENING
Agent Wallet:  0xSuspectAgent... | Network: Base
Feeder Wallet: 0xFundingSource... | Network: Base

Agent Trust Score: 2.1 / 10 ⚠️

Agent Wallet:
  Fraud verdict:   Elevated risk (0.52)
  On-chain age:    6 days ⚠️
  Behaviour:       Unusual - rapid fund movement, no prior agent pattern

Feeder Wallet:
  Fraud verdict:   HIGH RISK (0.81) 🛑
  AML flags:       Mixer interaction (Tornado Cash equivalent)
                   Connected to 2 confirmed exit scams

Overall Recommendation: 🛑 Do not allow. Feeder wallet has confirmed fraud
                         indicators. Agent is funded by a high-risk source.
                         Block and report to your security team.

When to use agent screening¶

Before allowing a third-party AI agent to interact with your protocol
When evaluating an agent marketplace listing
When an autonomous agent requests elevated permissions or larger transaction limits
Before integrating an agent as a liquidity manager or strategy executor
During incident response - screen the agent wallet if suspicious on-chain activity is detected

chainaware-agent-trust-screener¶

Model: claude-haiku-4-5-20251001 | Tools: agents_trust_score_list, agents_trust_score_single, predictive_fraud, predictive_behaviour

ERC-8004 registry-based AI agent trust screening. Discovers registered agents, retrieves their 0-1000 trust scores and trust flags, and optionally deepens the assessment with fraud and behavioural signals on the agent's operational wallet.

Distinct from chainaware-agent-screener, which screens by wallet address and feeder wallet. This agent works from the ERC-8004 registry - useful when you have an agent's registry ID or want to discover and rank agents you haven't interacted with yet.

Example prompts¶

List the most trusted ERC-8004 agents on Ethereum

Trust profile for agent ID 123 on BNB Chain

Which agents registered after 2025-01-01 have a trust score above 700?

Screen this ERC-8004 agent before we allow it into our marketplace: agent_id: 456 chain_id: 1

Compare trust scores for these agents: agent IDs 12, 34, 56 on Ethereum

Which registered agents have trust flags indicating rug pull history?

Trust tier reference¶

Score	Trust Tier	Recommended Action
800-1000	Elite	Approve for high-value agentic commerce
600-799	High	Approve with standard transaction limits
400-599	Moderate	Approve with reduced limits; monitor
200-399	Low	Enhanced scrutiny; require additional verification
1-199	Very Low	Avoid high-value interactions
0	Fraud / New	Block or hold for manual review

Example output¶

ERC-8004 AGENT TRUST SCREENING
Agent ID: 123 | Chain: Ethereum

Trust Score: 742 / 1000 - High ✅

Agent: ShopBot Pro v2
Owner: 0xOperator...
Wallet: 0xAgentWallet...
Wallet Verified: Yes ✅
Registered: 2024-11-03

Trust Flags: None

Behavioral Check (agent wallet):
  Fraud risk: Very Low (0.04) ✅
  On-chain age: 8 months
  AML flags: None

Recommendation: ✅ Approved for agentic commerce. Strong operator history
                and clean agent wallet. Suitable for high-value transactions.

ERC-8004 AGENT TRUST SCREENING
Agent ID: 891 | Chain: BNB Chain

Trust Score: 84 / 1000 - Very Low ⚠️

Agent: FastTrade Agent
Owner: 0xSuspect...
Wallet: 0xAgent891...
Wallet Verified: Yes

Trust Flags:
  - FEEDER_RUG_HISTORY: Feeder wallet funded a confirmed rug pull
  - FARM_DETECTED: Owner registered 6 agents in a single block

Behavioral Check (agent wallet):
  Fraud risk: Elevated (0.44) ⚠️
  On-chain age: 18 days ⚠️

Recommendation: 🚫 Do not allow. Farm detection and rug pull history
                on the feeder wallet. High probability of coordinated
                malicious activity.

When to use `chainaware-agent-trust-screener` vs `chainaware-agent-screener`¶

	chainaware-agent-screener	chainaware-agent-trust-screener
Input	Wallet address + feeder wallet	ERC-8004 agent_id + chain_id
Approach	Behavioral fraud + rug pull signals	Registry trust score + optional behavioral check
Best for	Any agent wallet (not necessarily ERC-8004 registered)	ERC-8004 registered agents; discovery and ranking
Trust score	0-10 derived from fraud + behaviour	0-1000 from ChainAware Agent Trust Score

Specialised Agents¶

chainaware-governance-screener¶

Example prompts¶

Governance tiers and multipliers¶

Example output¶

chainaware-gamefi-screener¶

Example prompts¶

Player verdicts¶

Experience tiers (for matchmaking)¶

Example output¶

chainaware-agent-screener¶

Example prompts¶

Trust score reference¶

Example output¶

When to use agent screening¶

chainaware-agent-trust-screener¶

Example prompts¶

Trust tier reference¶

Example output¶

When to use chainaware-agent-trust-screener vs chainaware-agent-screener¶

Further Reading¶

When to use `chainaware-agent-trust-screener` vs `chainaware-agent-screener`¶