ask-ast-mapper

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted source code from a repository.\n
  • Ingestion points: Source code files read by the tool in scripts/map_deps.sh.\n
  • Boundary markers: None present.\n
  • Capability inventory: Includes Read, Grep, Glob, and Subprocess.\n
  • Sanitization: None performed.\n- [COMMAND_EXECUTION]: The skill implementation contradicts its 'Strictly Read-Only' metadata. The scripts/map_deps.sh script uses output redirection (>) to write its results, enabling the agent to potentially overwrite files on the system if a path is provided to the second argument.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 04:33 PM
Security Audit — agent-trust-hub — ask-ast-mapper