The Agent Skills Directory

[METADATA_POISONING]: The skill frontmatter includes a risk: safe field, which is a deceptive claim intended to pre-emptively influence security assessments.
[EXTERNAL_DOWNLOADS]: The skill references an external third-party component, nanobanana-ppt-skills (Phase 3), which is not associated with any well-known service or trusted organization.
[INDIRECT_PROMPT_INJECTION]: The workflow involves processing untrusted external data during document conversion and spreadsheet automation phases, which could contain hidden instructions for the agent.
Ingestion points: Phase 2 (Import data), Phase 4 (Identify source format), and Phase 7 (Connect to data sources).
Boundary markers: None identified in the provided instructions.
Capability inventory: File organization (file-organizer), database integration (libreoffice-base), and automation of Google Workspace tools (gmail-automation, google-drive-automation).
Sanitization: No sanitization or validation steps are defined for processed data.

office-productivity