The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs a network request to a user-defined server (COOKIE_CLOUD_HOST) to download encrypted session data.
[CREDENTIALS_UNSAFE]: The skill requires and processes sensitive credentials, specifically a UUID and an encryption password. It instructs the agent to read these from environment variables or ask the user, providing mitigation advice against hardcoding secrets in source code.
[INDIRECT_PROMPT_INJECTION]: The skill represents an attack surface for indirect prompt injection as it ingests and processes data from an external, potentially attacker-controlled host.
Ingestion points: Reads encrypted JSON data from the URL {HOST}/get/{UUID}.
Boundary markers: None specified for the external data payload.
Capability inventory: The skill writes code to inject cookies into browser contexts (Playwright, Puppeteer, Selenium) and performs HTTP requests.
Sanitization: Relies on JSON.parse and local decryption, but does not explicitly describe sanitization of the decrypted session data before it is processed by the agent or injected into the project.

cookiecloud-sync