The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches community comment data from well-known platforms including GitHub, Reddit, and Bilibili via official APIs or automated browsing.
[COMMAND_EXECUTION]: Utilizes webfetch and agent-browser tools to collect data and interact with external web pages.
[PROMPT_INJECTION]: Vulnerable to indirect prompt injection because the skill ingests and processes untrusted user comments from public forums.
Ingestion points: Data scraped from external URLs is saved to comments_raw.json and passed to the agent for processing.
Boundary markers: The workflow description does not include the use of explicit delimiters or instructions to the agent to disregard embedded commands in the input data.
Capability inventory: The skill is capable of performing semantic clustering, argument extraction, and generating Markdown reports; it does not demonstrate system-level execution based on data content.
Sanitization: Pre-processing steps are defined for data quality (spam and bot removal) but do not include security-focused sanitization to prevent prompt injection attacks.

opinion-miner