The Agent Skills Directory

[SAFE]: The skill consists of documentation and usage examples for the Ollama command-line tool. It does not include executable code, scripts, or instructions that perform unauthorized actions.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing untrusted data (e.g., content from local files like ticket.txt) through an LLM classification model. While this creates a surface for indirect prompt injection, the scope is limited to local classification and does not lead to sensitive actions.
Ingestion points: Reads from local files or piped input via shell redirection (SKILL.md).
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the examples.
Capability inventory: The skill is limited to classification tasks (yes/no, multi-choice, scoring) using the local ollama tool; no file writing, networking, or dynamic execution capabilities are defined.
Sanitization: No sanitization of the input text is mentioned or performed in the provided examples.

ollama-decision