The problem RugSlash solves

RugSleuth helps crypto users quickly assess whether a Base token or wallet looks risky before they trust it.

Today, checking a suspicious token is slow and manual: users jump between BaseScan, liquidity trackers, deployer history, holder data, and social signals, then try to combine everything under time pressure. Most retail users skip this step, which makes them easier targets for rug pulls and low-effort scam launches.

RugSleuth turns that workflow into an autonomous investigation. A user pastes a Base address, and the agent buys the public data it needs from Apify Actors using x402 payments, analyzes the evidence, and returns a transparent rug-risk verdict with reasons and limitations.

It makes due diligence:

Faster: one address in, structured risk summary out.
Safer: users get warnings before interacting with suspicious tokens.
More transparent: every paid data request and evidence source is visible.
Agent-native: the AI agent can pay for external intelligence itself instead of relying on a prepaid API account.
Reusable: the same pattern can support token screening, protocol risk checks, wallet investigation, fraud monitoring, and other Web3 intelligence workflows.

Challenges we ran into

One major challenge was making the x402 flow work cleanly inside an autonomous agent workflow instead of as a simple CLI demo.

The agent does not call Apify directly. The flow is:

Codex agent -> custom MCP server -> x402 client -> Apify Actor
That meant we had to carefully separate responsibilities:

Codex decides what evidence it needs.
The MCP server exposes safe, structured tools.
The x402 client handles 402 Payment Required, signs the challenge, retries the request, and emits payment events.
The orchestrator streams everything back to the dashboard.