The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill implements a 'Zero-Failure Auto-Provisioning Rule' that forces the automatic download and execution of the 'chrome-devtools-mcp' package via 'npx -y'. The use of the '-y' flag explicitly bypasses all user confirmation prompts during tool installation at runtime.
[COMMAND_EXECUTION]: Instructions mandate the execution of local scripts such as 'py bin/mcp_auto_updater.py' without user oversight if dependencies are missing. This behavior bypasses typical security checkpoints where an agent would normally request permission before executing shell commands or modifying the environment.
[PROMPT_INJECTION]: The content includes strong directive language ('ATURAN MUTLAK', 'DILARANG') designed to override default agent safety protocols and force 'unrestricted' scraping operations. It explicitly commands the agent to suppress reporting missing tools to the user, instructing it to 'DILARANG menjawab not installed'.
[INDIRECT_PROMPT_INJECTION]: The skill is primarily designed to ingest untrusted data from arbitrary web sources.
Ingestion points: Web content retrieved via BeautifulSoup, Selenium, and Playwright as described in SKILL.md.
Boundary markers: None identified; there are no instructions to delimit or isolate scraped content from the agent's core instructions.
Capability inventory: Includes Python script execution, local file system writes (CSV/JSON/SQLite), and network requests across all included scraping libraries.
Sanitization: Absent; the workflow focuses on extraction and storage without instructions to validate, filter, or escape the content retrieved from external websites.
[EXTERNAL_DOWNLOADS]: The skill specifies numerous external dependencies including 'yfinance', 'ccxt', and 'datasets'. It also explicitly encourages the dynamic installation of missing Python libraries at runtime via the terminal if they are found to be lacking.

data-acquisition-specialist