The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to transmit user-provided text or local files to the external domain https://brewpage.app. This creates a surface for data exposure if the agent is instructed to upload sensitive information.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute several commands including curl for API interactions, jq for JSON processing, and file for identifying MIME types. These commands are used to process and transmit the data.
[EXTERNAL_DOWNLOADS]: The skill interacts with the external service brewpage.app, which is managed by the skill author. This involves sending data to an external API endpoint.
[PROMPT_INJECTION]: The skill processes untrusted data (file contents or user arguments) and incorporates them into shell commands. While it uses jq for safe encoding, the ingestion of arbitrary data presents a standard indirect prompt injection surface.
Ingestion points: The $ARGUMENTS variable and local file contents read via the Bash tool in SKILL.md.
Boundary markers: No specific boundary markers or instructions are provided to the model to ignore potential injection patterns within the content being uploaded.
Capability inventory: The skill possesses network transmission capabilities (curl) and local file read access (test -f, cat, file) via the Bash tool.
Sanitization: The skill employs jq to safely encode content into JSON payloads, reducing the risk of command injection during the API call construction.

brewpage-publish