With Ragmir, give your agents the cited context from your confidential documents.
The open-source local RAG for your AI agents. Index your confidential documents, repository files, and docs on your machine. Your agents get only the cited passages that matter, over MCP. Nothing leaves, nothing leaks.
A developer opens Codex, installs Ragmir, drops an aviation spec into .ragmir/raw/, indexes docs + the local file, then generates cited E2E tests.
Install
One package to activate queryable local documentation.
Ragmir starts as an open-source library and CLI. Install it where docs, runbooks, decisions, contracts, or questionnaires already live, then generate local state and agent helpers. The default mode stays offline, with no model to download.
Choose your package manager
Pick your package manager once. Every command below adapts to match.
Agent workflows
A local knowledge base for agents, onboarding, and teams.
Ragmir connects the same local project state to the agents you actually use. Cloud agents like Claude, Codex, Kimi, and GLM query the cited context; OpenCode stays the local runner option; Ollama is only for loading a local model when synthesis must stay on your machine.
Agent command
Target clients
Keep the categories explicit: cloud agents query Ragmir, OpenCode can run locally, and Ollama is the model runtime for fully local synthesis.
Cloud agents
Claude, Codex, Kimi, and GLM-style cloud agents are the same category here: they ask Ragmir for cited local context, then reason from the returned passages.
OpenCode
OpenCode is the local agent runner option when you want to keep the workflow close to the workstation and avoid another hosted RAG service.
Ollama
Ollama is not an agent target. It loads local models so the generation step can stay 100% local, confidential, and sovereign.
Use cases
What your agents and teams can do with Ragmir.
Point Ragmir at your confidential documents, repository files, and docs. Your agents get the cited passages relevant to the task, not a token-heavy dump of the whole corpus.
Docs & onboarding
Onboard from the docs, not from memory.
Index your cahier des charges, runbooks, ADRs, and decisions. The agent finds what to read before coding, and ships changes grounded in cited sources: onboarding a dev goes from weeks to minutes.
Specs & contracts
Change risky code with the full spec.
Before an agent refactors or migrates, Ragmir gives it the cited contracts, specs, and past decisions: changes are planned on what the documents actually say, not on a guess.
Runbooks & tracing
Trace an issue across your runbooks.
Ask where a flow is documented and why it fails; Ragmir returns the cited passages from your runbooks and incident notes, giving the agent the context it usually lacks.
Team memory
Answer without waking up the person who knows.
Team members query policies, decisions, incidents, and project notes from the same bounded local context.
Support handover
Give support the whole account history.
Index customer notes, implementation docs, tickets, and past decisions locally so an agent can draft answers or handover briefs with cited context.
Sales engineering
Answer RFPs and security questionnaires.
Retrieve evidence already written in security docs, contracts, SLAs, and compliance answers without copying it into a third-party SaaS.
Audit evidence
Prepare audits from the evidence you already have.
Point Ragmir at policies, controls, logs, and internal procedures so agents can assemble cited evidence packs without uploading regulated material.
Vendor due diligence
Explore a data room or a vendor dossier.
Analyze contracts, policies, technical appendices, and compliance documents with citations instead of scattered manual reading.
Legal and tax
Compare clauses, obligations, and evidence.
Contracts, tax notes, company memos, or residency research stay local while the agent retrieves the relevant sources.
Research and reports
Write from cited sources.
Briefings, Markdown reports, and summaries can separate evidence, inferences, uncertainties, and items to review.
Content & media
Automate a YouTube channel from a document corpus.
Feed Ragmir a shelf of books (psychology, history, any domain) and let your agent pull cited passages to draft video scripts, episode outlines, and talking points, all grounded in the source material instead of invented claims.
Audio and micro-learning
Listen to a dense dossier instead of reading it.
Generate a short audio summary (MP3/WAV) from cited passages with rgr audio, to review a spec, an architecture, or a watch without staying in front of the screen.
Sovereign and private
The key strength: make documentation useful without exporting it.
Ragmir keeps control at the project level: no telemetry, no automatic upload to a platform, no vague compliance promise, and explicit limits on what gets indexed.
Documents on disk
You point Ragmir at the relevant folder. Private files stay in the repository or the local workspace.
Git-ignored state
The index, reports, and generated helpers live in the local git-ignored .ragmir/ folder.
Redaction and explicit limits
Redaction, size limits, and auditing of unsupported files avoid claiming to index more than what is reliable.
Minimal logs
Access records stay metadata-oriented so usage can be checked without exposing sensitive content.
Honest scope
Where Ragmir shines, and where it does not.
Ragmir is a tool, not a silver bullet. It is built for specific, high-value scenarios. Use it where it helps, skip it where something else is better.
Ragmir is best at
- Confidential non-code corpora: PDFs, DOCX, XLSX, specs, contracts, cahier des charges that Claude Code cannot grep into.
- Corpora that exceed the model context window: when 200k tokens are not enough, retrieval is necessary, not optional.
- Strict confidentiality: legal, tax, client dossiers that must never leave your machine.
- Grounding AI agents in real evidence: cited passages with file and chunk references, traceable answers.
Less useful for
- Code search: Claude Code with ripgrep and native model understanding is already better on structured code.
- Small text corpora that fit in the context window: paste them directly, the model sees everything.
- Anything that needs cloud sync or multi-device real-time collaboration: Ragmir is local-first by design.
Desktop client coming
Desktop will come after the open-source core, with direct download.
The Tauri client should make the local flow visible: pick projects, watch folders, and run the same Ragmir commands without a terminal. No fake pricing page or waiting list until the native release is ready.
Follow the projectRagmir Desktop
PlannedDistribution planned through direct downloads and sideloadable installers, not through the App Store or Play Store.
Project registry
Each local folder keeps its own Ragmir state and its own access limits.
Direct distribution
Desktop installers and Android APK/sideload are the targeted paths before any store consideration.
Same local core
The app wraps Ragmir instead of creating a second hosted index or cloud document storage.
FAQ
Frequently asked questions
What Ragmir does, what it never sends off your machine, and how to connect it to your agents.
What is Ragmir?
Ragmir is an open-source library, CLI, and MCP server that turns your confidential document corpus (cahier des charges, specs, runbooks, corpora from Google Drive) into cited, queryable context for your AI agents. It indexes your documents on your machine and returns cited passages: your agents answer from your real documents instead of guessing. It is not an IDE code indexer; it complements it for non-code material. It works offline by default.
Does Ragmir send my confidential code or documents?
No. Ragmir only indexes the local folders you choose and keeps all generated state in the git-ignored .ragmir/ folder. No telemetry, no upload to the cloud, and offline by default: it fits confidential code, customer data, and regulated environments.
Which AI agents and tools does Ragmir work with?
Claude Code, Codex, Kimi, GLM/Zhipu-style cloud clients, OpenCode, and internal wrappers that can call a local CLI or helper. Run rgr install-agent for native helpers, or use rgr serve-mcp, the CLI, and the TypeScript library directly.
Does Ragmir require an API key, a model download, or an internet connection?
No. The default local-hash mode runs entirely offline, with no model or API key. Optional semantic embeddings (Transformers.js) can be preloaded once for better search quality, then also run offline. Generation stays outside Ragmir: you can hand citations to a cloud agent, or to a local model loaded with Ollama when the answer must stay on your machine.
How does Ragmir differ from an IDE's code indexing or a cloud RAG?
IDEs already index your code well. Ragmir targets what they cannot reach: your non-code documents (cahier des charges, specs, contracts, corpora from Google Drive). Unlike cloud RAG, Ragmir keeps everything on your machine, exposes it over MCP to any agent, and returns cited passages: local, cited, MCP, with no vendor lock-in.
Does Ragmir write the answers for me?
No. Ragmir returns cited passages from your local files; your agent or model writes from those sources. By keeping generation out of the core, answers stay grounded in real evidence and every citation is verifiable.
Which file formats can Ragmir index?
Source code, text, Markdown, PDF, Office and OpenDocument, EPUB, HTML, CSV, JSON, and YAML, plus the custom text extensions you enable. Unsupported files are flagged explicitly, and scanned PDFs can use an optional OCR command.
Is Ragmir free and open-source?
Yes. Ragmir is MIT-licensed and free, with no account required. The CLI, the library, and the MCP server are the current product; a desktop client is planned later as a direct download.
Install Ragmir, keep the dossier local, query the evidence.
No hosted account is required, and no data is exported to a third-party service.