openviking-skills
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill guides the agent in managing local skill assets, which includes directory management, creating zip archives for migration, and setting execution permissions (
chmod +x) on helper scripts within the~/.openviking/skills/directory. - [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of skills from remote Git repositories via the
add_skilltool. It includes a specific safety protocol to only install from external sources explicitly named by the user or after obtaining direct user confirmation. - [INDIRECT_PROMPT_INJECTION]: As a management tool that reads and processes external
SKILL.mdfiles from various AI platforms and repositories, it maintains an inherent surface for indirect prompt injection. The skill mitigates this by instructing the agent that external skill instructions do not override the user's primary instructions or safety boundaries. - [DYNAMIC_EXECUTION]: The skill supports the execution of helper scripts associated with managed skills. It also involves runtime generation of metadata (name/description) for local skills that lack standard frontmatter during the migration process.
- [SAFE]: The skill exhibits high security awareness by requiring the agent to scan files for secrets, credentials, and internal hostnames before allowing an upload. It also warns against putting credentials in
SKILL.mdfiles, particularly when sharing skills at the account level.
Audit Metadata