The Agent Skills Directory

[PROMPT_INJECTION]: The skill has a high attack surface for Indirect Prompt Injection (Category 8) because it retrieves raw data from public web platforms and passes it to AI-powered analysis tools.
Ingestion points: Data is ingested from X/Twitter (tweets, replies, timelines), Weibo, Bilibili, CSDN, WeChat articles, and Google search results via the fetch_tweet.py, fetch_china.py, and camofox_client.py scripts.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within the fetched content are defined in the documentation or provided to the agent.
Capability inventory: The skill includes an AI-powered profile analyzer (x-profile-analyzer.py) and various fetchers that interact with local and remote services, allowing malicious instructions in fetched content to potentially influence the agent's logic or downstream actions.
Sanitization: There is no evidence of content sanitization or validation to prevent control characters or injection patterns from reaching the AI model.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of 'Camofox,' an external browser service. Installation instructions involve cloning a repository from GitHub (github.com/jo-inc/camofox-browser) and installing dependencies via npm, which introduces external software into the execution environment.
[COMMAND_EXECUTION]: The skill provides instructions for setting up persistence using cron jobs (*/15 * * * * python3 tweet_growth_cli.py --run --fast) to automate monitoring tasks, which involves scheduled command execution.

x-tweet-fetcher