veto-implement-feature

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external spec files or feature descriptions supplied via arguments, which creates a potential surface for indirect prompt injection if those files contain malicious instructions meant to subvert the agent's logic.
  • Ingestion points: Reads content from user-specified file paths or plain-text descriptions provided in the '$ARGUMENTS' variable (SKILL.md).
  • Boundary markers: The skill lacks explicit delimiters or instructions for the agent to ignore potentially malicious embedded commands within the ingested specification data.
  • Capability inventory: The agent has the capability to execute shell commands through 'cargo test', 'make check', and 'make e2e' (SKILL.md).
  • Sanitization: There is no evidence of input validation or sanitization of the specification content before it is used to guide code generation and testing.
  • [DYNAMIC_EXECUTION]: The skill facilitates the runtime compilation and execution of code as part of its core functionality.
  • Evidence: Instructions direct the agent to automatically execute 'cargo test', 'make check', and 'make e2e', which involves compiling and running code modified or generated during the session.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 03:40 AM
Security Audit — agent-trust-hub — veto-implement-feature