The Agent Skills Directory

[SAFE]: The skill primarily consists of configuration files and instructions for model routing and agent behavior. No malicious patterns such as exfiltration, unauthorized code execution, or credential theft were detected.
[COMMAND_EXECUTION]: The skill includes a Python script (scripts/check-native-agent-profiles.py) intended to validate and install sub-agent profiles to a local directory. Analysis of the script confirms it performs integrity checks (SHA256) and uses standard library functions for file operations (shutil.copy2). It does not perform network requests or execute arbitrary shell commands.
[INDIRECT_PROMPT_INJECTION]: The agents-memory-updater sub-agent is designed to process conversation transcripts to extract user preferences and workspace facts. This creates a surface for indirect prompt injection from data in transcripts. However, the sub-agent instructions include explicit guardrails to filter out transient details and secrets, and the scope is restricted to updating a specific memory file (AGENTS.md).
[EXTERNAL_DOWNLOADS]: The installation instructions reference the author's official GitHub repository (TheAngryPit/TheAngrySkills). These references are consistent with standard skill distribution and do not involve untrusted third-party sources.

model-capability-router