The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses the sensitive file ~/.gooseworks/credentials.json to retrieve authentication tokens.
Evidence: Uses a python3 command to read api_key and api_base from the home directory to set environment variables for subsequent API calls.
[PROMPT_INJECTION]: The skill ingests untrusted data from TikTok, creating an indirect prompt injection surface where malicious instructions could be embedded in user-controlled content.
Ingestion points: TikTok profile bios, video titles, descriptions, and hashtag search results retrieved from api.gooseworks.ai or api.orth.sh.
Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded instructions when processing the external data.
Capability inventory: The skill can perform network requests using curl and execute shell commands.
Sanitization: Absent; no validation or escaping of the external content is performed before it is added to the agent context.
[COMMAND_EXECUTION]: The skill executes shell environment exports and python3 one-liners during setup to parse configuration files.

tiktok-search