content-hunter
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content (titles, summaries, and tags) from external social media platforms, which represents an attack surface where malicious data could influence the agent's behavior during summarization or reporting.
- Ingestion points: Extracts data from Xiaohongshu, Douyin, and Bilibili via
browser.snapshot()as specified in the execution flow (SKILL.md, Section 4.3). - Boundary markers: Absent; there are no specific instructions to the agent to treat scraped data as non-executable content or to use delimiters.
- Capability inventory: The skill has the capability to create directories, write files to the local workspace, and execute a Python script (
get_transcript.py) via the shell (SKILL.md, Sections 4.2, 4.5). - Sanitization: No sanitization or validation of the scraped content is mentioned before it is presented to the model for report generation.
- [COMMAND_EXECUTION]: The skill utilizes shell commands to manage its lifecycle and perform specific tasks.
- Schedules automated tasks using
openclaw cron add(README.md, Section 7.1). - Executes a local Python script
get_transcript.pywith video URL arguments to extract Bilibili subtitles (SKILL.md, Section 4.5).
Audit Metadata