skill-best-practices

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script (scripts/validate-metadata.py) to validate metadata inputs. This is a deterministic tool used for procedural validation within the skill-authoring workflow.
  • [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection by ingesting untrusted metadata and using it as arguments in a shell command. Ingestion points: metadata fields in SKILL.md Step 1. Boundary markers: absent in the command template. Capability inventory: subprocess call to python3 in SKILL.md. Sanitization: the validation script performs regex-based filtering on the provided arguments.
  • [DATA_EXPOSURE]: No access to sensitive files, environment variables, or credentials was observed. The skill operates entirely on the text provided for metadata validation.
  • [REMOTE_CODE_EXECUTION]: No remote script downloads, untrusted package installations, or remote code execution patterns were detected. All logic is contained within the local skill directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 12:15 AM