bencium-code-conventions

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow involving reading project files and executing commands, which creates a surface for instructions embedded in data to influence the agent.
  • Ingestion points: The skill instructs the agent to 'Read relevant files' and 'Explore' the project codebase during the development workflow.
  • Boundary markers: There are no explicit instructions for the agent to distinguish between trusted project instructions and potentially malicious external content embedded in the codebase files.
  • Capability inventory: The skill defines capabilities for command execution (test scripts, curl, typecheck), SQL execution in chunks, and file system modifications.
  • Sanitization: No sanitization, validation, or filtering of the content being read is requested.
  • [COMMAND_EXECUTION]: The instructions require the agent to execute shell commands for testing ('execute all until done'), performing typechecks, and verifying APIs via 'curl' commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 03:52 PM
Security Audit — agent-trust-hub — bencium-code-conventions