The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's primary function is to upload local files or inline content to a third-party hosting service at https://deed.page via curl. While this is the stated purpose, it inherently enables the movement of data from the agent's environment to an external public platform.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands, specifically curl for network transfers and tar for archiving directory contents before upload.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection. An attacker could provide malicious content (e.g., HTML with scripts or phishing elements) which the agent, following these instructions, would then host publicly on deed.page. The analysis identifies the following mandatory evidence for this category:
Ingestion points: The skill processes files or content provided by the user or generated by the agent for publication (e.g., report.html, ./site folder).
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the files being hosted.
Capability inventory: The skill uses curl for network requests and tar for file system operations in SKILL.md.
Sanitization: There are no explicit sanitization or validation steps for the content being uploaded, although the documentation notes that the remote service may reject content with 'executable or phishing markers'.

deed-page