xy-link
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local bash script (
scripts/bridge-skill.sh) to perform filesystem management tasks such as symlinking and directory verification. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided skill names and paths to generate bridge files (
SKILL.md) for the Grok agent, creating a surface for potential instruction injection if the inputs are maliciously crafted. \n - Ingestion points: The
inputargument passed tobridge-skill.sh(e.g., skill names or directory paths) inSKILL.md. \n - Boundary markers: No specific delimiters or "ignore instructions" warnings are included in the generated Grok bridge file content. \n
- Capability inventory: The script performs directory creation (
mkdir), symlinking (ln), file removal (rm), and file writing (cat) inscripts/bridge-skill.sh. \n - Sanitization: The script uses
basenameto extract names andpwd -Pfor path resolution, but it does not escape or sanitize the resulting strings before interpolating them into the Grok bridge file's frontmatter and body.
Audit Metadata