jaml
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill contains instructions to download components from the jaml-group/jaml-skills repository via npx in workflows/runtime.md.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external repository files and web pages, creating a surface for injection. Ingestion points: Repository files and retrieved pages (SKILL.md). Boundary markers: Explicit instructions to treat data as evidence, not commands (SKILL.md). Capability inventory: File system access and tool execution (workflows/runtime.md). Sanitization: Separation of inputs into model data (SKILL.md).
- [DYNAMIC_EXECUTION]: The skill handles JAML definitions and plugins which are considered executable code as per SKILL.md and workflows/compose.md.
- [COMMAND_EXECUTION]: Workflow instructions in workflows/runtime.md include a shell command for skill installation.
Audit Metadata