raglit

A local document RAG index you can stand up in one command. Point it at a folder, ask questions, or hand it to an agent as an MCP tool. One portable SQLite file is the whole index — pure-Go, single static binary, no services to run.

Quickstart

# build
go install github.com/iodesystems/raglit/cmd/raglit@latest

# 1. one-time setup — pick an OpenAI-compatible endpoint + models (a wizard)
raglit init

# 2. ingest a folder (code, markdown, text; PDFs get OCR'd)
raglit ingest ./my-project --now

# 3. ask
raglit search "how does the auth token refresh work?"

That's it. init asks for a base URL + API key and lists the endpoint's models so you can pick a chat/vision model and an embedding model; everything else uses sensible defaults (you never pass model flags again). The index lives in ~/local/raglit by default.

No endpoint handy? Every offline piece works without one: raglit demo runs a self-contained tour, and text ingest falls back to a dependency-free splitter when no model is configured.

What it does

• Ingest folders, files, or URLs (file://, http(s)://) — lazily (queued) or with --now. Text/code is segmented by the model into coherent, ~500-word fragments (functions bound with their docs, not atomised); PDFs are OCR'd page by page with cross-page stitching.

• Search — BM25 (--mode bm25, default), vectors (--mode vec), or hybrid RRF (--mode hybrid). Results are precise citations: document → page → fragment.

• Serve — expose the index(es) to any MCP client (Claude Desktop, agentkit):

  raglit serve

  Tools: search, ingest, index_status, list_indexes.

Commands

raglit init                          configure endpoint + models (wizard)
raglit ingest TARGET... [--now]      queue folders / files / URLs (lazy; --now drains)
raglit search "query" [--mode M]     M = bm25 | vec | hybrid
raglit status                        documents/fragments, queue progress, rate, ETAs
raglit serve                         stdio MCP server
raglit demo                          offline, self-contained tour

--home DIR picks the index home; --index NAME selects a named index within it. With no --index, commands use the config's default_index (set in the wizard), falling back to default.

Daemon mode

For big or ongoing ingests, run a daemon that owns the index and the background workers; other invocations call into it over HTTP instead of touching the SQLite files (no contention, and remote-capable):

raglit daemon --addr 127.0.0.1:7420        # long-running: workers + HTTP API

# point any command at it (or set RAGLIT_DAEMON=http://host:7420)
raglit ingest --daemon http://127.0.0.1:7420 ./my-project
raglit search --daemon http://127.0.0.1:7420 "rollback procedure"
raglit status --daemon http://127.0.0.1:7420

Endpoints: GET /health, GET /indexes, POST /ingest, GET /search, GET /status. The daemon reads local paths from its OWN filesystem (share it, or use http(s):// targets); bind to localhost — there's no auth yet.

For agents (agentkit)

raglit's search output is exactly the shape agentkit's ragnotify.ParseHits consumes, so one raglit serve drives both a model's explicit searches and agentkit's proactive "live-watch" pings (the finder scopes which indexes it watches via Opts.ExtraArgs {"index": ...} — all by default). raglit's agent.DocFinder also plugs straight into a Session's FinderPreparer.

How it's built

• SQLite FTS5 gives BM25 + the document:page:fragment index in one pure-Go dependency (modernc.org/sqlite, no CGo). Vectors are stored as BLOBs, cosine brute-forced (fine for a local corpus).
• LLM segmentation (via agentkit) turns pages/text into coherent fragments with a schema-validated fix-loop and a safe fallback; an open fragment is carried across page/window boundaries and embedded only once it's resolved.
• Multi-index — one home holds several named indexes; serve searches all (RRF-merged, tagged) or a scoped subset.

Home layout

~/local/raglit/            (or $RAGLIT_HOME, or --home)
  config.json              endpoint + model settings (from `raglit init`)
  index.sqlite             the default index (index-<name>.sqlite for others)
  originals/               copies of ingested sources
  pages/                   page images for OCR

Everything for an index is under one directory — copy it, back it up, or delete it wholesale.

Roadmap

• ◻ Daemon auth + remote file upload (today: localhost, shared-FS or URL targets).
• ◻ Vector reranking; opt-in summaries for oversized fragments.