T09 · Insecure Skill Coding Practices
Error
- Location
index.ts:362- Finding
AMM Operations Disable Slippage Protection
- Content
View full analysis
- Remediation
View remediation
Security audit
Agent Market
Security checks for vulnerabilities and agentic risk
Overview
The skill is a coherent Base Sepolia prediction-market integration, but it can immediately sign wallet transactions and has weak trade-safety controls.
Review before installing. Use only a dedicated Base Sepolia wallet with limited test funds, never a reused private key, and manually confirm every write transaction. Be especially careful with buy, sell, and add-liquidity actions because the skill sets minimum received amounts to zero. Avoid automating HEARTBEAT.md arbitrator actions unless there is a separate human approval step.
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)
Vulnerability Patterns
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (7)
Missing User Warnings
High
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
The oracle governance functions can approve bond transfers, propose outcomes, dispute proposals, finalize markets, arbitrate disputes, and reset oracle state immediately on invocation. In a prediction-market skill, these are especially dangerous because they can both spend funds and directly determine market resolution, so prompt injection or accidental invocation could cause financial loss and governance manipulation.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The skill exposes users to real blockchain transactions involving USDC, trading losses, oracle bonds, and irreversible settlement actions, but it does not prominently warn that operations are final, can lose funds, and may require signed on-chain transactions. In this context, the absence of explicit risk and confirmation guidance increases the chance that an agent or end user will execute financially harmful actions without informed consent.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The market creation function directly performs ERC20 approval and then submits a live createMarket transaction as soon as it is invoked. In an agent skill context, that means a prompt-influenced call can immediately spend wallet funds and create an irreversible on-chain position without any built-in preview, confirmation gate, or policy checks.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The buy functions execute token approval and swap transactions immediately, with slippage protection hardcoded to 0n, meaning any amount of shares is accepted. In an autonomous or semi-autonomous agent, this creates a real risk of prompt-triggered fund loss, MEV/front-running exploitation, or severe price execution without the user being warned before funds are committed.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The sell functions immediately approve outcome tokens and submit live sell transactions with minimum output set to 0n. This can cause users to liquidate positions at arbitrarily poor prices if the market is manipulated or front-run, and there is no in-code confirmation barrier to prevent an agent from being socially engineered into selling holdings.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The liquidity functions approve funds and alter LP positions immediately without any explicit user warning or confirmation. Adding or removing liquidity changes portfolio exposure and may strand funds in illiquid or adversarial markets, which is particularly risky when actions can be triggered through agent workflows rather than a dedicated wallet UI.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 89% confidence
- Finding
These functions invoke irreversible on-chain actions such as claiming, emergency withdrawal, and post-resolution withdrawal directly when called. While some are beneficial, they still affect wallet state and can realize losses, lock in outcomes, or interact with untrusted contracts without a final user authorization step in the skill itself.
- Content
Static analysis
No suspicious patterns detected.