solana-development
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install core Solana development tools and frameworks from official and well-known repositories, including the Solana toolchain installer and the Anchor framework.
- [COMMAND_EXECUTION]: Instructions include standard development commands using cargo, anchor, npm, and the solana CLI to initialize projects, build programs, and run tests.
- [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is auditing and testing Solana programs, which involves the ingestion of user-provided code files. This presents a surface for indirect prompt injection if the audited code contains malicious instructions targeting the agent.
- Ingestion points: Program source files (.rs), configuration files (Cargo.toml, Anchor.toml), and client scripts (.js, .ts) provided by the user (as referenced in SKILL.md and references/testing-practices.md).
- Boundary markers: The skill mitigates this risk by providing extensive security checklists and validation patterns designed to identify and isolate malicious logic during the auditing process.
- Capability inventory: Compilation and testing capabilities provided via cargo, anchor, and npm CLI tools (referenced in references/anchor.md and references/testing-frameworks.md).
- Sanitization: The skill relies on structured security review processes and specific vulnerability detection patterns to distinguish between executable code and potential injection attempts.
Audit Metadata