The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted content from external URLs, tweets, and local files, which introduces a surface for indirect prompt injection where malicious instructions embedded in the content could influence the agent's behavior.
Ingestion points: Content is ingested via tools such as WebFetch, fetch_tweet, and Jina Reader, as well as direct local file reads mentioned in SKILL.md.
Boundary markers: The prompt instructions lack specific delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: The skill's capabilities are focused on structured analysis and report generation; no dangerous capabilities like subprocess execution or network write operations are defined in the skill files.
Sanitization: There are no sanitization steps described to validate or escape external content before analysis.

analyzer