solana-sniper-bot
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection through unverified blockchain data.\n
- Ingestion points: Token mint addresses and account metadata are fetched from the Solana blockchain and Raydium API within
scripts/sniper.py.\n - Boundary markers: The prompt used in the
llm_evaluatefunction interpolates this external data into a string without delimiters or instructions to ignore embedded commands.\n - Capability inventory: The LLM's assessment results directly trigger financial transactions (token swaps) via the Jupiter aggregator, using the user's private key.\n
- Sanitization: No sanitization or validation is performed on the token metadata strings before they are incorporated into the LLM prompt.
Audit Metadata