The Agent Skills Directory

[DATA_EXPOSURE] (SAFE): The scripts/ai_model_validator.py script accesses local files like wrangler.toml and constellation.json. However, it only performs keyword checks to verify if specific service bindings are present and does not extract secrets or transmit data to external servers.
[COMMAND_EXECUTION] (SAFE): No dangerous command execution patterns (e.g., os.system, subprocess.run with untrusted input) were detected. The Python script is limited to basic file path checks and text reading.
[PROMPT_INJECTION] (SAFE): The markdown content in SKILL.md contains instructional content for users but lacks any patterns associated with bypassing safety filters, system prompt extraction, or 'Ignore previous instructions' attacks.
[EXTERNAL_DOWNLOADS] (SAFE): The skill does not attempt to download external scripts, binary executables, or unverified packages from the internet.

emerging-tech