Convert documents into private agent-ready Markdown with doc7
Use a local or private multimodal model to convert PDFs, Office files, scans, screenshots, charts, formulas, and diagrams into Markdown for agent search and reasoning.
npx skills add agentskillexchange/skills --skill convert-documents-into-private-agent-ready-markdown-with-doc7
doc7 setup to select a local or private OpenAI-compatible vision model endpoint, then convert files with doc7 or register the MCP server with command /absolute/path/to/doc7 and args ["mcp"].Use doc7 when an agent needs document content transformed into searchable, quotable Markdown without sending files to a hosted document-processing service. The operator workflow is concrete: install the doc7 CLI, point it at a local or private OpenAI-compatible vision model, convert files or directories directly, or expose its MCP `convert_to_markdown` tool to an agent runtime. Invoke this instead of using doc7 manually when the job is repeated agent ingestion, RAG preparation, evidence extraction, or document triage where the agent needs structured Markdown plus conversion metadata. The boundary is private multimodal document-to-Markdown conversion for agent workflows; it is not a generic document AI platform, OCR SDK, or broad CLI listing.
What this skill actually does
Inputs and prerequisites: doc7 CLI, local or private OpenAI-compatible multimodal endpoint such as LM Studio or Ollama, optional MCP-compatible client.
Setup notes: Install the released CLI using the repository’s checksum-verifying installer or a release archive, run doc7 setup to select a local or private OpenAI-compatible vision model endpoint, then convert files with doc7 or register the MCP server with command /absolute/path/to/doc7 and args [“mcp”] .
Source and verification boundary: use https://github.com/magicrew/doc7 as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a MCP workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.