deepwork

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a set of procedural instructions for an AI agent to act as a scheduler and orchestrator. It does not contain any executable scripts, binary files, or external dependencies.
  • [DATA_EXPOSURE]: The skill manages its own persistent state by writing to the .slim/deepwork/ directory and explicitly instructs the agent to update .gitignore and .ignore to handle these files. This is a standard project management practice to keep local state files out of version control while remaining accessible to the agent tools.
  • [COMMAND_EXECUTION]: The skill defines a workflow that invokes other specialized internal agents (such as @oracle, @librarian, @explorer, and @designer). These references are consistent with standard multi-agent orchestration and do not involve arbitrary command execution or privilege escalation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:19 AM
Security Audit — agent-trust-hub — deepwork