ai-skill-manager
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a shell script (
scripts/validate_all_skills.sh) and documentation for executing local Python scripts (quick_validate.py,package_skill.py) to perform validation and packaging of other skill directories. These commands are intended for local repository maintenance. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and audit external
SKILL.mdfiles as part of its cataloging and routing workflows. - Ingestion points: Recursively scans directory structures (using
find) to identify and process the content of allSKILL.mdfiles in a given repository. - Boundary markers: None explicitly specified for the input processing; the skill relies on the structure of the YAML frontmatter and markdown body of the analyzed files.
- Capability inventory: The skill instructions and associated shell script allow for the execution of Python and Bash commands on the host system.
- Sanitization: The skill does not implement its own sanitization logic for ingested markdown, instead directing the content to specialized validation scripts.
- [SAFE]: The identified data ingestion and command execution patterns are functional requirements for a management and validation utility. No evidence of data exfiltration, obfuscation, or unauthorized remote code execution was found.
Audit Metadata