rust-skill-creator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text data by browsing or fetching documentation from external Rust registries.\n
- Ingestion points: External URLs from docs.rs or doc.rust-lang.org are ingested into the context via agent-browser or WebFetch.\n
- Boundary markers: Absent. The instructions do not define distinct encapsulation boundaries or filters to prevent untrusted content from being interpreted as instructions.\n
- Capability inventory: File system write capabilities via local file system creation (mkdir, file generation in ~/.claude/skills/).\n
- Sanitization: Absent. The content retrieved from the web block is directly interpolated into new skill metadata templates.\n- [DYNAMIC_EXECUTION]: The skill generates new SKILL.md documentation and reference files dynamically at runtime based on predefined structural templates and user arguments.\n- [COMMAND_EXECUTION]: Executes shell and tool-specific commands such as mkdir, ls, and custom plugin routines to scaffold new local skill directories.
Audit Metadata