The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a Python script (scripts/send_message.py) that can be executed from the CLI to send messages. This is the primary intended function of the skill.
[CREDENTIALS_UNSAFE]: The skill follows secure practices by instructing users to store their POKE_API_KEY in an environment variable (~/.zshrc) rather than hardcoding it. The script retrieves this key via os.environ.get().
[DATA_EXFILTRATION]: The skill performs network operations to poke.com. This is an expected behavior for a notification service and is documented as the skill's purpose. It does not attempt to access sensitive files like SSH keys or AWS credentials.
[EXTERNAL_DOWNLOADS]: The skill communicates with the official API endpoint https://poke.com/api/v1/inbound-sms/webhook. This is a well-known service domain related to the skill's purpose.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests data via the -m CLI flag or through stdin (file: scripts/send_message.py).
Boundary markers: None explicitly defined in the script, as it treats the input as a raw string message.
Capability inventory: The script has network access (urllib.request.urlopen) to a specific vendor URL.
Sanitization: The input is JSON-encoded before being sent to the API, preventing basic protocol injection, though the content itself is sent as-is.

poke-assistant