solana-sniper-bot
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a setup script (
scripts/setup.sh) that executes shell commands to install several Python dependencies required for blockchain interaction and network communication. - [DATA_EXFILTRATION]: The skill manages a sensitive
SOLANA_PRIVATE_KEYretrieved from the environment (typically configured in a.envfile) to sign blockchain transactions. It performs legitimate network operations to external services, including the Anthropic API (a trusted service), and the Raydium and Jupiter APIs to monitor liquidity pools and execute swaps. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the processing of untrusted blockchain metadata.
- Ingestion points: External token metadata, such as mint addresses and holder concentration, is fetched from the Raydium API and Solana RPC nodes in
scripts/sniper.py. - Boundary markers: The skill lacks boundary markers or instructions to ignore embedded commands when interpolating this metadata into the LLM evaluation prompt.
- Capability inventory: The bot possesses the capability to execute automated financial transactions (swaps) based on the output of the LLM risk assessment.
- Sanitization: There is no evidence of sanitization or validation of the fetched metadata before it is included in the prompt, potentially allowing an attacker to influence the bot's risk scoring through malicious token attributes.
Audit Metadata