project-workspace
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill manages file synchronization between local and remote environments. It includes a specific security guideline instructing the agent not to include keys, tokens, or private local information in instructions files before uploading.
- [COMMAND_EXECUTION]: The skill describes file management operations such as moving and archiving directories. These operations are guarded by instructions to verify sources and targets and to seek user authorization before destructive actions (moving or overwriting).
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for indirect prompt injection by ingesting and processing project files (e.g., AGENTS.md, PROJECT_RULES.md) which may originate from external sources.
- Ingestion points: Reads content from local and remote project configuration files via file-system access and the
filebrowser-skill. - Boundary markers: Instructions require the agent to compare file differences and report them to the user rather than performing silent overwrites.
- Capability inventory: Performs directory creation, file movement (archiving), and network synchronization via external tools.
- Sanitization: The skill relies on mandatory human-in-the-loop confirmation and explicit user authorization for all synchronization and archival tasks.
Audit Metadata