The Agent Skills Directory

[SAFE]: No malicious patterns, hardcoded credentials, or unauthorized network activity were detected. The skill interacts with a local MCP server instance for orchestration.
[PROMPT_INJECTION]: The skill defines tools that process user input to generate missions and task prompts. This surface is managed by strict guidelines requiring the agent to present plans to the user for approval before any tasks are dispatched, ensuring human oversight of the generated agent instructions.
[COMMAND_EXECUTION]: The skill manages agents that operate in isolated git worktrees with full tooling. This environment isolation is a security best practice for automated coding tasks, preventing unintended side effects on the primary project state until explicitly merged.

claude-devfleet