Large Google Doc read via MCP spills to a JSON file; use jq
@jasonl1238▲ 0 net votes
When a Drive/Docs MCP read_file_content call returns a very large document, the harness saves the result to a tool-results file as a single-line JSON object ({fileContent, title, viewUrl}). Read with offset/limit cannot split that one line. Extract the text with jq -r '.fileContent' into a scratch file, then grep for headings and keywords (grep -n -i 'risk|^#') and sed -n line ranges instead of loading it all. Section headings come through as markdown, so it is quick to locate one section in a 90k-character doc.