Growth & Marketing Agents¶

Ten agents that use on-chain behavioural prediction to drive user acquisition, personalisation, retention, and revenue growth for Web3 platforms.

Setup required: For AI Agents — MCP registration and agent installation.

chainaware-onboarding-router¶

Role: Routes new users to the right onboarding flow based on their on-chain behavioural profile.

What it does: Evaluates a connecting wallet's behavioural signals (predictive_behaviour) and fraud risk (predictive_fraud) to classify the user and route them to the most appropriate onboarding experience. A DeFi power user gets the advanced flow; a first-time user gets the guided flow; a flagged wallet gets a compliance hold.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-onboarding-router Route this new user for onboarding: wallet 0xNEW...USER

Output includes:
- User behavioural classification (Power DeFi / Active Trader / Casual / New User)
- Fraud risk check
- Recommended onboarding flow
- Personalisation signals (interests, experience level, likely use cases)

chainaware-whale-detector¶

Role: Identifies high-value wallet holders and VIP users for priority treatment.

What it does: Analyses behavioural signals to detect whale wallets — users with large holdings, high transaction volumes, or deep DeFi engagement. Combines predictive_behaviour (activity signals) and predictive_fraud (legitimacy check) to distinguish genuine high-value users from wash traders or inflated wallets.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-whale-detector Is wallet 0xBIG...HOLDER a genuine whale worth VIP treatment?

Output includes:
- Whale classification (Mega Whale / Whale / Large Holder / Standard)
- Legitimacy assessment (genuine vs. inflated/wash)
- Value signals (portfolio size indicators, DeFi depth, tenure)
- Recommended treatment (VIP programme, account manager outreach, standard)

chainaware-defi-advisor¶

Role: Recommends DeFi products and strategies personalised to a wallet's behavioural profile.

What it does: Reads a wallet's DeFi engagement history via predictive_behaviour and fraud standing via predictive_fraud, then generates personalised product recommendations — yield strategies, lending opportunities, liquidity pool suggestions — matched to the user's demonstrated risk appetite and activity patterns.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-defi-advisor What DeFi products should we recommend to wallet 0xDEFI...USER?

Output includes:
- User's DeFi profile (risk appetite, preferred protocols, activity patterns)
- Personalised product recommendations (3–5 suggestions)
- Rationale for each recommendation
- Products to avoid based on profile
- Engagement hook for each recommendation

chainaware-airdrop-screener¶

Role: Screens airdrop recipients to eliminate Sybil wallets and low-quality targets.

What it does: Evaluates candidate wallets for airdrop eligibility by combining behavioural authenticity signals (predictive_behaviour — genuine engagement vs. Sybil farming patterns) with fraud risk (predictive_fraud). Produces a tiered eligibility list that rewards genuine users and filters out airdrop farmers.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-airdrop-screener Screen these wallets for airdrop eligibility: [0xA, 0xB, 0xC, 0xD]

Output includes:
- Per-wallet eligibility tier (Tier 1 / Tier 2 / Tier 3 / Ineligible)
- Sybil risk flags
- Fraud risk flags
- Recommended allocation multiplier per tier
- Summary statistics (eligible %, flagged %)

chainaware-cohort-analyzer¶

Role: Segments a set of wallets into behavioural cohorts for targeted campaigns and product decisions.

What it does: Takes a list of wallet addresses and analyses each using predictive_behaviour and predictive_fraud, then clusters the results into meaningful user segments — power users, casual holders, churned users, high-value prospects, at-risk users. Suitable for campaign planning, feature prioritisation, and retention strategy.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Sonnet 4.6 (complex cohort reasoning across multiple wallets)

Example invocation:

@chainaware-cohort-analyzer Segment these 20 wallets into cohorts: [0x1, 0x2, ... 0x20]

Output includes:
- Cohort assignments per wallet
- Cohort profiles with key characteristics
- Size and composition of each cohort
- Recommended actions per cohort (retain, upsell, re-engage, monitor)
- Campaign targeting suggestions

chainaware-lead-scorer¶

Role: Scores inbound leads (wallet connects or sign-ups) by conversion potential.

What it does: Evaluates a wallet's behavioural signals to predict likelihood of conversion, upsell, or long-term retention. Uses predictive_behaviour to assess engagement depth and predictive_fraud to filter out low-quality leads. Outputs a lead score and recommended next action for sales or marketing automation.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-lead-scorer Score this inbound lead: wallet 0xLEAD...PROSPECT

Output includes:
- Lead score (0–100)
- Lead quality tier (Hot / Warm / Cold / Disqualified)
- Key buying signals
- Fraud / bot risk check
- Recommended next action (demo outreach / nurture / standard onboarding / disqualify)

chainaware-upsell-advisor¶

Role: Identifies upgrade and upsell opportunities based on a user's current behaviour and platform usage.

What it does: Analyses existing user wallets to identify signals indicating readiness for an upsell — increased transaction volume, new DeFi activity, engagement with premium features. Combines predictive_behaviour (growth signals) and predictive_fraud (legitimacy check) to prioritise upsell outreach.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-upsell-advisor Which of these users are ready for upsell? [0xUSER1, 0xUSER2, 0xUSER3]

Output includes:
- Upsell readiness score per wallet
- Specific trigger signals (activity increases, feature usage)
- Recommended product or tier to upsell to
- Suggested messaging angle
- Priority ranking for outreach

chainaware-platform-greeter¶

Role: Generates a personalised welcome message for a connecting wallet based on their on-chain history.

What it does: When a wallet connects to a platform, this agent reads their behavioural profile and fraud standing to generate a contextually appropriate greeting. A returning DeFi power user gets acknowledgement of their history; a new user gets an introductory message; a flagged wallet triggers an internal alert instead.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Haiku 4.5

Example invocation:

@chainaware-platform-greeter Generate a welcome message for connecting wallet 0xRETURNING...USER

Output includes:
- Personalised greeting text (ready to display in UI)
- User profile summary (for internal use)
- Fraud risk check result
- Recommended UI experience adjustments

chainaware-marketing-director¶

Role: Orchestrates multi-wallet marketing analysis and campaign strategy using agent capabilities.

What it does: A higher-level orchestration agent that uses Claude's agent tool to coordinate complex marketing tasks — analysing large user cohorts, generating campaign briefs, producing segmentation strategies, and synthesising growth recommendations. Calls predictive_fraud as part of quality filtering during orchestration.

Tools used: predictive_fraud (via Agent orchestration for complex multi-step tasks)
Model: Claude Sonnet 4.6 (orchestration and strategic reasoning)

Example invocation:

@chainaware-marketing-director Analyse our user base of 500 wallets and produce a Q2 campaign strategy

Output includes:
- User base segmentation summary
- Campaign brief per segment
- Channel recommendations
- Budget allocation suggestions
- KPIs and measurement framework
- Prioritised action list

chainaware-wallet-marketer¶

Role: Creates wallet-personalised marketing content and outreach for individual users.

What it does: Takes a target wallet address and produces personalised marketing content — email copy, in-app message, push notification — calibrated to that wallet's specific on-chain behaviour, interests, and risk profile. Uses predictive_behaviour for personalisation signals and predictive_fraud to ensure the target is a genuine user.

Tools used: predictive_behaviour, predictive_fraud
Model: Claude Sonnet 4.6 (content generation with nuance)

Example invocation:

@chainaware-wallet-marketer Create a personalised re-engagement email for wallet 0xCHURNED...USER

Output includes:
- Personalised subject line
- Email / message body copy
- Personalisation rationale (which signals drove the content)
- A/B variant suggestions
- Recommended send timing based on activity patterns