The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from multiple external sources, which could lead to instructions being processed as data.
Ingestion points: SKILL.md identifies channels such as emails, messenger services (WhatsApp, Telegram, Teams), and document uploads.
Boundary markers: The instructions do not specify the use of delimiters or specific prompts to isolate external content from agent instructions.
Capability inventory: The skill utilizes tools like kanzlei-allgemein-bea-journal for ZIP extraction and journal management, and kanzlei-allgemein-output-versand for dispatching results.
Sanitization: The skill advises against assuming the authenticity of messenger inputs and suggests contextual verification.
[COMMAND_EXECUTION]: The workflow involves instructions for handling and decompressing external data packages.
Evidence: SKILL.md directs the agent to export messages as ZIP files and unpack them for assignment to case files.

intake