git-worktree
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to perform various shell and git commands to manipulate the filesystem and repository. Evidence includes the use of
git rev-parse,git worktree add,basename, anddirnamewithin conditional shell logic inSKILL.md.- [INDIRECT_PROMPT_INJECTION]: The skill constructs shell commands using a 'slug' derived from untrusted user input, creating a surface for indirect prompt injection. - Ingestion points: User task descriptions or requests.
- Boundary markers: Absent. The shell script template in
SKILL.mddoes not include escaping or quoting for the<task-slug>placeholder. - Capability inventory: Full shell command execution via the agent's command tools.
- Sanitization: Absent. The skill provides a natural language instruction to use 'ASCII kebab-case,' but lacks programmatic validation or sanitization in the script logic.- [DYNAMIC_EXECUTION]: The skill dynamically generates and executes shell scripts to manage worktree state. The 'Default Workflow' section in
SKILL.mdprovides a multi-line shell script template that the agent is expected to customize with user-influenced variables and execute at runtime.
Audit Metadata