The Agent Skills Directory

[DATA_EXFILTRATION]: The skill transmits task metadata, including the machine name, project name, task status, and result summary, to the external Bark push notification service at https://api.day.app. This is the primary intended function of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes data from external files and agent-generated text without strict sanitization.
Ingestion points: Project names are extracted from the AGENTS.md file, and task summaries are provided via command-line arguments.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: The script send_bark_notification.py has network access to perform POST requests using Python's urllib.request library.
Sanitization: Input data is URL-encoded for the API request, but there is no validation or filtering of the actual content to prevent injection patterns.
[PERSISTENCE]: The user documentation provides instructions for manually adding environment variables to shell initialization files (~/.bashrc or ~/.zshrc). This is a standard method for persisting configuration but falls under the category of modifying system shell profiles.

bark-notify