The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources which presents a potential attack surface.
Ingestion points: The skill uses the web_fetch tool and processes user-supplied files or pasted text as described in SKILL.md.
Boundary markers: It includes a specific security instruction in the 'Untrusted content' section to treat all external content as data only, ignoring any embedded instructions or commands found within those sources.
Capability inventory: The agent has access to web search and fetch tools, as well as standard shell and file system operations to execute its tasks.
Sanitization: The skill mandates a 4-tier source hierarchy for verifying facts and includes a 'no-harmful-output' rule to prevent commercial or legal damage, which acts as a secondary layer of sanitization for ingested content.

prompt-improver