workspace-isolation
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands to manage Git worktrees and run project-specific build and test tools (e.g., npm, cargo, pip, go).
- [EXTERNAL_DOWNLOADS]: It triggers downloads of project dependencies from official registries via standard package managers (npm, pip, cargo, poetry, go) based on the presence of manifest files.
- [PROMPT_INJECTION]: The skill ingests and follows instructions found in local project files such as
CLAUDE.mdto determine directory preferences, creating a surface for indirect prompt injection. - Ingestion points:
CLAUDE.md,package.json,Cargo.toml,requirements.txt,pyproject.toml,go.mod(SKILL.md) - Boundary markers: Absent; the skill is instructed to follow
CLAUDE.mdpreferences "without asking." - Capability inventory: Executes shell commands for Git operations and various build/test tools (SKILL.md).
- Sanitization: Absent; configuration strings from
CLAUDE.mdand branch names are used directly in shell commands.
Audit Metadata