Questions about IngestMD

IngestMD turns any codebase or document folder into clean, AI-ready Markdown, then gives an AI model the right slice of it instead of the whole tree. It runs as a local-first desktop app, a command-line tool, and an MCP server, all over a shared Rust core.

Repomix and GitIngest flatten a repository into one file fast, which IngestMD also does. The difference is the decision layer on top: describe a task and IngestMD ranks, dependency-expands, and packs the exact files the model needs to a token budget, discovers the real subsystems in a codebase, builds diff review bundles, and ingests documents, all local by default.

Yes. The free tier is a full deterministic bundler with secret redaction and binary safety that are never paywalled, .gitignore respect, token counts, the MCP bundle tool, and Collection export. Pro and Team add intelligent task context, diff review, area discovery, document ingestion, and governance.

No. Scanning, bundling, redaction, and area discovery run on your machine, offline, with no account and no telemetry. Network egress is opt-in only: a local LLM you run yourself, optional Pro cloud models with your own key kept in the OS keychain, and Confluence retrieval when you ask for it.

IngestMD ships a built-in MCP server, so it connects to Claude Code, Cursor, VS Code, and Codex over the Model Context Protocol. You can also export Markdown bundles to paste into any AI chat, or send a token-budgeted context to your own local LLM.

macOS (Apple Silicon and Intel), Windows, and Linux. The desktop app is built with Tauri and Rust; the macOS builds are signed and notarized by Apple.

The free bundler respects .gitignore, skips binaries and build output, and takes your include and exclude globs and stack templates. Pro task context goes further: it ranks files by relevance to a task you describe, expands their dependencies, skeletonizes large files, and packs the result to a token budget with a reason for each file.

Yes. Area discovery clusters a repository into its real subsystems by imports and shared terms, one titled, token-budgeted bundle per area, so you can hand the AI just the subsystem you are working on. It has been measured across 21 large repositories including Linux, Node, VS Code, PyTorch, and Kubernetes.

Yes. It converts PDF, Word, and Excel files into clean Markdown with a pure-Rust pipeline, and retrieves Confluence Cloud pages and spaces. Documents are redacted the same way code is.

Yes, by default and never paywalled. API keys, tokens, and private keys are scrubbed before anything is written, and a NUL-byte content guard plus binary extension filtering keep binaries out. Redaction applies to documents too.

Yes. Salesforce mode groups Apex, LWC, and objects into the features they form by following @salesforce/apex calls and SOQL, instead of handing the AI a flat folder of metadata.

The free tier is available now and stays free. Pro is planned at $9 per month or $99 for a lifetime license, and Team at $19 per user per month, adding shared profiles, CI generation of context bundles, and an audit log of what was sent to AI. See the pricing section on the home page for the current details.