skill-system-foundry
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The provided Python scripts (such as scaffold.py, bundle.py, and validate_skill.py) execute standard file system operations, including file reading, writing, directory creation, and zip archiving. These operations are consistent with the skill's stated purpose of managing a skill system. The scripts utilize realpath-based boundary checks to prevent path traversal and follow a 'standard library only' constraint.
- [INDIRECT_PROMPT_INJECTION]: As a validation and auditing tool, the skill is designed to ingest and analyze untrusted data (other skill files). While this creates a theoretical attack surface, the scripts use deterministic logic for parsing and validation rather than direct prompt interpolation, and the skill does not provide a mechanism for external data to override the agent's core instructions.
- [DYNAMIC_CONTEXT_INJECTION]: The skill documentation (references/claude-code-extensions.md) describes the dynamic context injection syntax (
!command) for specific platforms as a feature. This is documented for educational purposes and is not used within the skill's own operational logic to execute arbitrary commands.
Audit Metadata