Skills
skills/majiayu000/claude-skill-registry/a2a-executor-patterns/Gen Agent Trust Hub

a2a-executor-patterns

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or behaviors were detected in the skill's logic or metadata. The provided patterns follow standard software development practices for building robust executors.\n- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection as it is designed to process external task data.\n
  • Ingestion points: Untrusted data enters via the A2ATask parameters in the executeTask pattern in SKILL.md.\n
  • Boundary markers: The implementation templates include a validateTask function as a defensive boundary.\n
  • Capability inventory: The skill is configured with Read, Write, Bash, Grep, and Glob tools.\n
  • Sanitization: The provided patterns include placeholders for validateParameters to ensure data is sanitized before execution.\n- [COMMAND_EXECUTION]: The skill provides instructions for using the Bash tool to run local maintenance and validation scripts, such as scripts/validate-executor.sh, which is consistent with its role as a development tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 03:03 PM