The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill leverages ScrapeGraphAI's ScriptCreatorGraph to synthesize standalone Python and Playwright automation scripts dynamically based on user prompts. These scripts are intended to be saved and executed within the user's workspace to perform specialized scraping tasks.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external business websites and Google Maps search results to extract structured lead data.
Ingestion points: Web content is retrieved from Google Maps search results and standalone business domains discovered during the process (specified in SKILL.md and references/workflow-architecture.md).
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore embedded instructions when the LLM parses the scraped DOM content.
Capability inventory: The agent has capabilities to execute Python code, write files (CSV, JSON), and perform network operations for scraping (specified in SKILL.md and references/data-schema.md).
Sanitization: While the skill sanitizes data for display by removing font icon glyphs and newline characters, it does not include specific defenses against prompt injection attacks embedded within the ingested web data.
[COMMAND_EXECUTION]: The skill requires the execution of shell commands for environment setup, specifically pip install for Python dependencies and playwright install for necessary browser binaries.
[EXTERNAL_DOWNLOADS]: The skill instructions involve downloading and installing official packages such as scrapegraphai and playwright from standard registries (NPM and PyPI), as well as Chromium browser binaries managed by Playwright.

google-maps-lead-generation