ellmos-skill-creator
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform local command-line operations as part of the development workflow, including running metadata validation scripts like
testing/skill_frontmatter_gate.pyand performing version control operations viagit commiton specific directory paths. - [EXTERNAL_DOWNLOADS]: The instructions reference external source code repositories for the
grounding-seedandsource-resolvermodules on GitHub. These are documented as infrastructure dependencies within the ecosystem. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a generator for code and documentation based on user-provided requirements, creating a surface for indirect prompt injection where untrusted input is interpolated into system templates.
- Ingestion points: User-provided skill intents and descriptions in
SKILL.md. - Boundary markers: The skill utilizes structured YAML frontmatter and JSON templates to maintain separation between instructions and data.
- Capability inventory: File system writes (creating new skill directories), subprocess calls (
python,git), and model tool invocations. - Sanitization: The skill relies on structural validation against JSON schemas to ensure output integrity.
Audit Metadata