The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an attack surface for Indirect Prompt Injection (Category 8). Since the primary purpose of the skill is to audit and analyze user-provided smart contract code (Rust/Anchor), a malicious actor could embed instructions within the code being analyzed to attempt to influence the agent's behavior or exfiltrate analysis results.
Ingestion points: The skill ingests untrusted code provided by the user via file input or prompt text during auditing tasks.
Boundary markers: While the skill provides a structured report format, it does not explicitly define strict delimiters or instruction-bypass protections for the ingested code.
Capability inventory: The skill involves deep analysis of logic, which could be steered if an attacker crafts code that mimics administrative instructions.
Sanitization: There are no explicit instructions for the agent to sanitize or escape the content of the smart contracts before processing them.
[EXTERNAL_DOWNLOADS]: The skill references numerous official documentation sites, GitHub repositories, and security resources related to the Solana ecosystem (e.g., solana.com, anchor-lang.com, spl.solana.com). These references are informative and point to well-known technology organizations and community-trusted services, which is consistent with the skill's educational purpose.

solana-security