custom-skills-tool-overlap-analyzer

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from project files (Skills, Agents, Commands, and Workflows) to perform "Deep Analysis" and semantic comparison.
  • Ingestion points: Reads content from files found in directories such as .claude/agents/, agents/, skills/, and commands/ (documented in SKILL.md).
  • Boundary markers: The workflow does not specify the use of delimiters or instructions to ignore embedded prompts when analyzing these external files.
  • Capability inventory: Includes file system traversal, reading file content, executing grep searches, and writing report files to the docs/report/ directory.
  • Sanitization: No evidence of sanitization or escaping for the ingested content is provided before it is evaluated by the agent.
  • [COMMAND_EXECUTION]: The skill uses the grep utility to scan the project for references to tool names in markdown, YAML, and JSON files as part of its cross-reference analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 08:16 AM
Security Audit — agent-trust-hub — custom-skills-tool-overlap-analyzer