The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill processes information from meeting transcripts, emails, and internal messaging, which are considered untrusted external data sources. This presents a potential surface where content within those communications could attempt to influence the agent's behavior. The skill addresses this by prioritizing structured signal normalization and requiring evidence-backed interpretations as outlined in SKILL.md.
[Local Configuration Path Access]: To coordinate automation features, the skill inspects local file paths such as ~/.codex/automations/ to check for existing monitoring records. This allows the agent to provide accurate setup guidance and prevent the creation of redundant automated tasks.

analyze-account-signals