Back to skill

Security audit

Play AgentDraw

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because it promotes daily wallet transactions as free while omitting network fees and clear consent for recurring activity.

Keep wallet confirmation enabled and authorize daily check-ins separately with a fee limit. Calls advertised as free can still consume BNB for gas, and cumulative fees may exceed rewards. Verify contract addresses and simulate transactions before signing. Ticket purchases can lose money. Live manifests and deployed contract claims remain unverified.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 4)May include surrounding context.

md
---
name: play-agent-draw
version: 1.2.0
description: Play AgentDraw — the daily on-chain draw on BNB Chain for Binance .ai Agent ID holders. START FREE: a companion check-in contract lets your wallet check in once per UTC day at value 0 (checkIn(), no approval), earning 1 point per day; 100 points redeem for exactly 0.0005 BNB via redeem(). The draw itself has NO NFT: buy tickets with enter(n) at exactly 0.0005 BNB per ticket (1–20 per transaction); 65% of every ticket forms that UTC day's prize pool and one winning ticket takes 100% of the pool at the permissionless settleDay() after 00:00 UTC. There is also an NFT companion, the AgentDraw Pass: mintPlayer() mints the free Player Pass at value 0 once the wallet holds 1+ check-in points, and mintFounder() mints the Founder Pass at exactly 0.001 BNB, half of it forwarded to the check-in pool inside the mint. Use when your human wants to do the free daily check-in, mint either Pass, enter today's AgentDraw, check today's pool or their t

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 88)May include surrounding context.

md
Agent ID, both open from day one, max 1 per wallet PER LANE (a wallet
may hold one of each; the mint flags never reset on transfer).

### Mint the FREE Player Pass (value 0, no approval)

Pre-flight, for YOUR wallet — stop if any check fails:

The skill instructs mainnet transactions advertised as free without warning that users still pay BNB gas fees.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The skill describes the mainnet check-in as costing the user nothing, but value 0 only means no BNB is transferred as the call's value; a standard wallet transaction still incurs gas unless explicitly sponsored. Encouraging daily check-ins without disclosing those fees can cause unexpected recurring expenses that may exceed the eventual redemption reward; transaction simulation does not correct the misleading cost claim.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 46)May include surrounding context.

md
- https://agentdraw.click/llms.txt
- https://agentdraw.click/draw.json

## Do this FIRST — the free daily check-in (value 0, no approval)

On the AgentDrawCheckIn contract (address from draw.json):