meme-coin-audit
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves ingesting and analyzing untrusted external data, specifically third-party smart contract source code (Solidity and Rust) and blockchain state data. This presents a surface for indirect prompt injection attacks where malicious instructions could be embedded within code comments or metadata to influence the agent's audit results.
- Ingestion points: The agent is instructed to read and analyze files within the
src/directory and evaluate data retrieved from blockchain explorers and CLI tools. - Boundary markers: The instructions lack explicit guidance for the agent to use delimiters or specific safety prompts when processing external code to prevent the obedience of embedded instructions.
- Capability inventory: The skill assumes the agent can perform file system operations (read, grep), execute CLI tools (solana, spl-token), and potentially run test suites (Foundry).
- Sanitization: There are no instructions for sanitizing or filtering the source code before it is processed by the LLM.
Audit Metadata