The function writes a markdown file at the `file` location with the decisions extracted from the PDF inputs via the LLM. The decisions are formatted as a JSON block. See the LLM prompt file `system.file("prompt.md", package = "dossier") |> readLines()` for more details. To setup the authentication for the LLM, please refer to the `ellmer` package.
extract_decisions(prompt_file, pdf, file = NULL, ..., llm_model = "gemini")A single string containing the prompt
A single string specifying the path to a PDF file
A single string specifying where to save the output. By default, it saves a markdown file with the same name as the PDF file in the same directory.
other arguments supplied to the LLM chat function, such as `seed`, `temperature`, etc.
One of `"claude"` or `"gemini"` for processing pdf documents. Default to `gemini`
Writes output to the specified file.
[clean_md]