grepai-trace-graph
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates tools that analyze local source code to construct dependency graphs. This creates an indirect prompt injection surface where malicious instructions embedded within the source code being analyzed (such as within function names or metadata) could potentially be interpreted by the agent.
- Ingestion points: The
grepai trace graphcommand reads function names and call relationships from source files (SKILL.md). - Boundary markers: The tool output uses structured tree or JSON formats, though it does not explicitly instruct the agent to ignore embedded natural language instructions within those structures.
- Capability inventory: The skill utilizes shell execution for
grepai,python3, and file redirection. - Sanitization: No specific sanitization or filtering of the analyzed code content is mentioned.
- [DYNAMIC_EXECUTION]: The skill provides examples of using Python heredocs (
python3 << 'EOF') to dynamically generate and execute scripts that transform JSON output into DOT or Mermaid diagram formats. - Evidence: Examples provided in the 'Visualizing Graphs' section of SKILL.md.
- Note: These patterns use static templates for data formatting and do not incorporate untrusted external inputs into the executable logic.
Audit Metadata