The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external, untrusted sources which can serve as a vector for malicious instructions.
Ingestion points: The scripts/recover_page.py script fetches raw HTML and Markdown content from archive.org, archive.today (multiple domains), and r.jina.ai based on user-provided URLs.
Capability inventory: The agent, after retrieving this content, typically processes it for research, summarization, or further tool execution, creating a path for embedded instructions to influence the agent's behavior.
Sanitization: While the script performs basic validation for content quality (rejecting small bodies or common interstitial titles like "Just a moment"), it does not sanitize or filter the content for malicious prompt injection patterns.
Boundary markers: The skill does not wrap the recovered content in safety delimiters or provide "ignore instructions" warnings before presenting the data to the agent's context.
[EXTERNAL_DOWNLOADS]: The skill fetches data from several external services to fulfill its purpose.
Services: The skill communicates with archive.org, archive.ph, archive.md, archive.li, archive.is, and r.jina.ai.
Functionality: It retrieves archived snapshots and server-side rendered versions of pages to bypass WAFs, paywalls, and bot detection.

blocked-page-recovery