The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/invoke.py uses subprocess.run to call a local management engine. This is a standard architectural component for this vendor to integrate the skill with the agent harness.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data retrieved from the Postmark API, such as message bodies and templates, which could contain malicious instructions. Ingestion points: Data from API responses in scripts/postmark.py is rendered into the agent's context. Boundary markers: The SKILL.md file provides explicit instructions for the agent to treat Postmark data as inert information and never as instructions. Capability inventory: The skill has network access (urllib.request in scripts/postmark.py) and local script execution capabilities. Sanitization: The skill relies on the LLM's adherence to prompt-based guardrails to prevent data from being interpreted as instructions.
[CREDENTIALS_UNSAFE]: The skill manages sensitive API tokens stored at ~/.kntnt/postmark/credentials.json. While this is a sensitive file path, the skill implements security best practices by enforcing 0600 permissions and using a clipboard-based setup to avoid exposing tokens in the conversation history.

postmark