The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and process data from local configuration files (lp.json and locale-specific JSON) and to scrape content from external live websites to extract attribution information (specifically targeting the .navara-attr-item class). This ingestion of external, untrusted content presents an indirect prompt injection surface.
Ingestion points: Data is read from docs/src/data/lp.json, docs/src/data/lp-locales/*.json, and external websites during the attribution scraping process.
Boundary markers: The instructions do not define specific delimiters or guardrails to prevent the agent from interpreting instructions that might be embedded in the scraped content or local data files.
Capability inventory: The skill environment allows for shell command execution via pnpm (for building, formatting, and linting), file system modifications, and network requests to fetch API data and scrape external sites.
Sanitization: There are no explicit instructions for sanitizing or validating the content retrieved from external websites before it is incorporated into the project metadata or locale strings.
[EXTERNAL_DOWNLOADS]: The skill fetches geographic data from the Overpass API using the well-known endpoints overpass-api.de and overpass.kumi.systems. These are standard sources for OpenStreetMap data used in the project's rendering workflow.
[COMMAND_EXECUTION]: The skill utilizes several local development commands via pnpm, including pnpm build, pnpm preview, pnpm run build:example, pnpm run format, and pnpm run lint. It also involves launching a browser via Playwright for automated screenshot capture as part of the verification process.

navara-lp