skill-best-practices
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill includes instructions to review existing SKILL.md files, which creates an indirect prompt injection surface where instructions in the analyzed data could influence agent behavior.
- Ingestion points: The agent is directed to read and evaluate SKILL.md files within the repository during review and troubleshooting workflows.
- Boundary markers: The skill does not explicitly define delimiters or specialized instructions to isolate content from analyzed skills during the review process.
- Capability inventory: The skill allows for local script execution (validate-skills.sh) and repository file modification (CLAUDE.md, README.md).
- Sanitization: The bundled validation script correctly uses yaml.safe_load in its Python component to mitigate risks when parsing YAML metadata from untrusted files.
- [COMMAND_EXECUTION]: The skill utilizes a local bash script (validate-skills.sh) to perform metadata validation, employing python3 and git to check repository state and file integrity.
- [EXTERNAL_DOWNLOADS]: When the --remote flag is used with the validation script, the skill invokes npx to fetch and run the 'skills' package from the NPM registry to verify skill availability.
Audit Metadata