SKILL /

agent skill

Complete reference for autonomous token risk scoring on Base.

What is con+?

Con+ is a token risk intelligence skill for Conway-compatible AI agents. It gives your agent the ability to assess rug pull risk before any on-chain interaction — swapping, providing liquidity, or deploying capital.

The underlying models are trained on 41,424 token pairs with 1,905 confirmed ground-truth rug pulls, achieving 74–91% precision depending on the risk tier.

How it Works

1.
Skill activation
Your agent detects a token interaction. It calls the con+ skill with the token address.
2.
Risk analysis
Con+ runs 3 ML models (Static + Activity + Hybrid) and returns a calibrated risk score within ~200ms.
3.
Tier decision
The agent receives a CRITICAL / HIGH / ALERT / WATCH / SAFE classification and decides to proceed, monitor, or abort.
4.
Autonomous action
For CRITICAL and HIGH tiers, the agent can automatically reject the transaction. For WATCH, it can proceed with reduced position size.

Quick Start

1. Add con+ MCP to your agent:
npx conplus-mcp
2. Your agent can now call con+ as a tool:
Example agent prompt
Before swapping any token, always call conplus_analyze
with the token address. If the tier is CRITICAL or HIGH,
abort the transaction immediately. If ALERT, reduce
position size to 10% of intended. Log all results.
3. REST API alternative:
curl
curl -X POST https://conplus.app/v1/risk/analyze \
  -H "Content-Type: application/json" \
  -d '{"address":"0x...", "chain":"base"}'

Risk Tiers

CRITICALNever interact. High confidence rug within 24h.
HIGHDo not ape. Strong drain signals detected.
ALERTCaution. Reduce size. Set stop-loss.
WATCHMonitor. Baseline elevated. Normal small position ok.
SAFEBelow threshold. Standard due diligence still applies.