The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python script scripts/recover_page.py and provides curl examples intended to be used by the agent to perform web requests and process content. The script is implemented using only Python's standard library.
[EXTERNAL_DOWNLOADS]: The skill fetches content from well-known external services including the Wayback Machine (archive.org), archive.today variants, and Jina Reader (r.jina.ai). These operations are consistent with the skill's stated purpose of content retrieval and do not involve executing untrusted remote code.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external web pages, creating a potential surface for indirect prompt injection if the agent interprets content within retrieved pages as instructions.
Ingestion points: scripts/recover_page.py fetches raw content from third-party URLs using urllib.request.
Boundary markers: The script provides visual separators (e.g., --- body ---) and metadata headers in its output to help the agent distinguish retrieved content from system instructions.
Capability inventory: The agent executes the local Python script and reads its output; the script itself is restricted to standard library network operations and does not have persistent or elevated permissions.
Sanitization: The skill identifies and rejects common interstitial or error pages (e.g., 'Just a moment', 'Access Denied') through title and content heuristics, reducing the likelihood of processing bot-protection pages.

blocked-page-recovery