Skip to content
AI HOT (Curated Pool)

Using LLMs in Script Shebang Lines

在脚本的shebang行中使用LLM

Simon Willison demonstrates using an LLM command in a script shebang line, with fragments generating SVG, the -T option calling llm_time, and a YAML template defining Python tools to compute 2344×5252+134 and return 12,310,822.

Why it matters: Simon Willison shows how putting the llm command in a shebang line turns a plain text file into an executable script, treating natural language as an interpreter for command-line automation.

Read the original ↗Export Markdown