The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands for diagnostics and maintenance, including npm run doctor, npm run build, and npx playwright install chromium.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading external resources, specifically browser binaries via Playwright and package updates using npm install -g awwwards-mcp@latest.
[DYNAMIC_EXECUTION]: The skill supports self-modification and runtime updates:
It encourages the agent to modify its own source code (src/parsers.ts) and update fixtures to fix 'parser drift'.
It identifies the environment variable AWWWARDS_AUTO_UPDATE=1 which enables a background self-update mechanism on startup.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from an external source (awwwards.com) and possesses the following security profile:
Ingestion points: Reads live web markup and JSON blobs from awwwards.com to diagnose and fix parsers.
Boundary markers: None identified in the provided instructions; the agent is encouraged to directly search and incorporate findings into source code.
Capability inventory: The skill has access to shell execution (npm), file system writes (modifying src/, test/, and scripts/), and network operations (fetching live pages).
Sanitization: No explicit sanitization or validation of the ingested markup is described before the agent uses it to re-anchor parsers.

awwwards-doctor