The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has a potential attack surface as it processes external content from Instagram post captions and user-supplied bio text which could contain hidden instructions.
Ingestion points: Reads profile descriptions, captions via list_channel_posts, and a local configuration file at ~/.uplika/<workspace-slug>/voice.md.
Boundary markers: The instructions do not define specific delimiters for the untrusted content.
Capability inventory: The skill is limited to reading data via list_accounts and list_channel_posts and generating text. It lacks high-risk capabilities like writing to the filesystem or executing shell commands.
Sanitization: No specific sanitization or filtering of the ingested content is described.
[SAFE]: The skill exhibits legitimate behavior consistent with its stated purpose. All external links and local file paths (e.g., uplika.com, ~/.uplika/) are resources owned or managed by the vendor. No evidence of malicious obfuscation, privilege escalation, or persistence was detected.

ig-profile