clarity-audit
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data by reading Clarity smart contract source files (.clar) through the --source parameter. This creates a surface for indirect prompt injection if the contract files contain malicious instructions meant to influence the agent's reasoning. \n
- Ingestion points: The .clar source files processed by the audit, quick-check, and function-review subcommands as defined in SKILL.md and AGENT.md. \n
- Boundary markers: The documentation does not specify the use of delimiters or clear instructions to the agent to ignore embedded instructions within the contract source code. \n
- Capability inventory: The clarity-audit-agent is designed to perform open-ended reasoning, deep dives, and security reviews based on the ingested contract code. \n
- Sanitization: There is no evidence of sanitization, filtering, or escaping of input data before it is processed by the agent. \n- [COMMAND_EXECUTION]: The skill documentation outlines a CLI interface that executes a local TypeScript file (clarity-audit/clarity-audit.ts) using the Bun runtime. This represents the execution of local shell commands. \n- [EXTERNAL_DOWNLOADS]: The SKILL.md file contains a reference link to an external GitHub repository (github.com/pbtc21/publisher-succession/issues/1) used to illustrate audit output examples.
Audit Metadata