Back to skill

Security audit

SelfPaid - Zero-Cost DeFi Payments

Security checks for vulnerabilities and agentic risk

Overview

Review recommended: the payment validator accepts excessive collateral deposits, and the advertised recovery times conflict with the skill’s own formula.

Keep automatic signing disabled until collateral limits are independently enforced and confirmations reflect decoded transaction amounts. Review total spending and the disclosed fee minimum, treat recovery estimates as uncertain, and pin any optional MCP installation to a reviewed release. The hosted operator receives payment details.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
validate.js
Finding

Remote transaction builder controls collateral spending beyond the approved amount

Content
View full analysis

Impact

Signing the accepted batch can move substantially more liquid USDC into a yield protocol than authorized, exposing those additional funds to protocol and liquidity risks. The position remains credited to the payer; this finding does not establish theft or attacker ownership.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

The payment workflow promises full capital recovery and unrestricted redemption without warning about DeFi financial risks.

High
Category
Output Handling
Confidence
99% confidence
Finding

The unconditional '100% recovery' and 'Zero Lockup' claims can induce autonomous agents to allocate funds without accounting for variable yields, protocol losses, USDC depegging, or withdrawal liquidity constraints. The examples also materially understate recovery time: a 77.25 USDC deposit at 4.58% APY takes roughly 2,600 days to earn 25.125 USDC under the documented simple-yield formula, not 230 days. Transaction validation cannot guarantee returns or redemption availability.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y selfpaid-mcp'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The recommended npx -y selfpaid-mcp command can download and execute an unpinned package version, automatically accepting installation without ensuring that the code matches a reviewed release. A compromised publisher account or malicious future release could execute code with the agent host's permissions and expose accessible credentials or manipulate transaction-building workflows. The financial automation context increases the potential consequences, although this README provides no evidence that the current package is malicious.

Content

No source excerpt is available for this finding.

The workflow sends payment details to a hosted service without a privacy warning.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The build request discloses the sender, recipient, amount, and collateral preferences to a hosted operator before execution, without explaining retention or obtaining specific privacy consent. Although executed payments become public on-chain, the request additionally exposes pending or abandoned payment intent and enables correlation with request metadata; HTTPS and local signing do not prevent the operator from observing it.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 101)May include surrounding context.

Get current APYs and see which protocol is currently offering higher yields:

bash
curl -s https://selfpaid.cryptofolio.it/api/v1/payment/rates

Response:

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 172)May include surrounding context.

Query live on-chain balances, active yield collateral (aBasUSDC / maUSDC), and accrued yield interest for any wallet:

bash
curl -s "https://selfpaid.cryptofolio.it/api/v1/payment/treasury?wallet=0x1111111111111111111111111111111111111111"

Response:

MCP server referenced without pinned version: 'npx -y selfpaid-mcp'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Running npx -y selfpaid-mcp automatically downloads and executes an unpinned package version, allowing a compromised release or publisher account to introduce arbitrary code. Execution inside a payment agent's environment could expose credentials or permit transaction manipulation; the documented calldata checks do not sandbox the MCP process.

Content

No source excerpt is available for this finding.

The validator enforces an undisclosed minimum developer fee rather than the manifest's stated 0.5% fee.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The validator requires a minimum developer fee of 0.1 USDC, exceeding the advertised 0.5% for payments below 20 USDC; a 1 USDC payment incurs a 10% fee. Because this amount is enforced on the developer-wallet transfer, autonomous agents relying on the advertised percentage can authorize unexpectedly high cumulative charges, particularly for micropayments. The code does not establish malicious intent, and the claim of eventual yield reimbursement does not eliminate the immediate excess charge.

Content

No source excerpt is available for this finding.