crypto-meme-scout
Fail
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill implements a workflow that fetches HTML content from an external website (dexscreener.com) and pipes it through a text processor (sed) directly into a Node.js process for execution. This allows any JavaScript embedded in the website's data variables to be executed on the host machine.\n- [DYNAMIC_EXECUTION]: The skill uses the Node.js vm module (vm.runInNewContext) to evaluate a string constructed from the scraped web data. This is an unsafe practice that treats untrusted external data as executable code, bypassing standard security boundaries.\n- [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes data from an untrusted source without sanitization. Malicious text on the website could trick the AI agent into performing unauthorized actions.\n
- Ingestion points: Data is ingested from dexscreener.com via curl in the SKILL.md file.\n
- Boundary markers: None are present to separate instructions from data.\n
- Capability inventory: Shell command execution via curl, sed, and node.\n
- Sanitization: None; the data is executed directly as JavaScript.\n- [COMMAND_EXECUTION]: The skill relies on complex shell pipelines to manipulate data, which is a high-risk pattern that increases the local system's attack surface.
Recommendations
- HIGH: Downloads and executes remote code from: https://dexscreener.com/?rankBy=trendingScoreH24&order=desc - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata