tiktok-topic-aweme-list
Warn
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The script
scripts/fetch_topic_aweme_list.pyreads sensitive shell configuration files, including~/.zshrc,~/.bashrc,~/.bash_profile, and~/.profile, to search for and extract theREDFOX_API_KEYcredential.\n- [DATA_EXFILTRATION]: The skill transmits the resolved API key and user query data to the external domainredfox.hk. The network implementation in the Python script explicitly disables SSL hostname validation (context.check_hostname = False), which bypasses a standard security check intended to prevent man-in-the-middle attacks.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from the TikTok platform (via a third-party API) and renders it into the conversation, creating a surface for indirect prompt injection.\n - Ingestion points: The
scripts/fetch_topic_aweme_list.pyscript fetches video content summaries, usernames, and metadata from theredfox.hkAPI.\n - Boundary markers: The instructions do not define clear delimiters or "ignore instructions" warnings for the agent when processing this external data.\n
- Capability inventory: The skill uses a Python script to perform network requests and format output into Markdown tables.\n
- Sanitization: The script includes basic sanitization (
escape_md) to handle newlines and pipe characters within the Markdown output.\n- [COMMAND_EXECUTION]: The skill's primary functionality is triggered by executing thescripts/fetch_topic_aweme_list.pyPython script via the command line.
Audit Metadata