The Agent Skills Directory

[COMMAND_EXECUTION]: Multiple scripts, including funding_detector.py, notifier.py, and monitor.py, utilize subprocess.run to interact with the host system. This is used to retrieve API credentials via the pass utility and execute local shell scripts for notifications.
[EXTERNAL_DOWNLOADS]: The monitoring logic performs network requests to various external sources. Scripts such as funding_detector_v2.py and scraper_free.py fetch data from the Firecrawl API and scrape information from corporate and news platforms like Tianyancha, 36kr, and Sina Finance.
[PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface through its data ingestion and analysis pipeline.
Ingestion points: External news snippets and business change records collected by the scrapers are processed for analysis.
Boundary markers: The analyze_with_llm function in scripts/analyzer.py interpolates external content into a prompt without using explicit delimiters or instructions to ignore embedded commands.
Capability inventory: The skill has broad tool permissions, including Bash, Exec, and Browser, which increases the potential impact if a malicious injection is processed.
Sanitization: External content is not sanitized or escaped before being included in the prompt for LLM evaluation.

healthcare-monitor