Magpie Search

A federated search engine — the search engine an AI agent or LLM reaches for when it needs to find something true to reason over.

Ever had your computer reboot on you, or a power outage hit mid-session? Every thread your agent was holding — gone. Now you have the tool to get it back. Never forget what your agent lost again. Magpie indexes everything your AI has ever worked through, locally, so a crash is a hiccup instead of amnesia.

What Magpie is

A normal search engine looks in one place. Magpie takes one question and fans it across everything that matters at once — the AI's entire conversation history, the files on the machine, a structured knowledge graph, a vector store, and the live web — and pulls the answer back from wherever it actually lives. Five sources, one call.

And it searches each one the right way. It can grep for an exact string or regex when you know the precise token — a file path, an error, a line of code. It can search by keyword. It can search by meaning, so it finds the thing even when the words don't match. It can do all of that at once.

Then it does the part that makes it trustworthy: it fuses everything into a single ranked answer, and every result carries a trust tier — fact > reference > lead > stale. The solid sources rise, the loose ones are marked as leads to verify, duplicates collapse, and it's all trimmed to fit so it never floods the AI's context. Ask it to go deep and it expands one question into many, reads the pages, and tells you how many independent sources agree — a full research sweep without an army of agents.

It runs entirely on the machine. No server, no account, and no telemetry unless you turn it on. The AI's transcripts and files never leave. It plugs into whatever AI is running over MCP, so the agent can reach all six sources the instant it needs them.

It is a tool for an AI — an agent or an LLM.

What's inside

At its core is a local index of the AI's transcripts: a SQLite database with two structures built side by side —

• an FTS5 full-text index (BM25 keyword ranking), and
• a vector index (sqlite-vec) of 384-dim embeddings produced locally by a small all-MiniLM-L6-v2 model.

Everything is redacted at ingest — a scrubber strips ~30 classes of secrets (keys, tokens, private keys, connection strings) before a single byte hits the index.

On top of that index sit the five search modes:

Around that sits the federation layer — the part that makes it federated:

• A provider plugin system. Five backends (transcripts, files, knowledge graph, vector, web), each returns Hit objects tagged with a trust tier.
• A fan-out: one query goes to all providers concurrently (≤8 workers), each with a 5-second timeout that fails open — a slow source contributes nothing rather than blocking the call.
• Trust-weighted RRF fusion — Reciprocal Rank Fusion where each source's rank is multiplied by its trust weight (fact ×3, reference ×2, lead ×1, stale ×0.3), damping constant 60. This is the math that merges six heterogeneous sources into one honest ranking.
• Cross-source dedup by content hash — the same fact found in three places collapses to one hit, tagged with where else it appeared (corroboration).
• A token-budget trim, so the merged set never overflows the calling AI's context.

And it exposes all of this to an AI over an MCP server — the tools it hands an agent are exactly: search, recent, session, list_sessions, stats, reindex. Note what's not in that list: nothing that writes an answer.

Why that is not RAG

RAG = Retrieval-Augmented Generation. It's a two-stage pipeline, and the defining stage is the second one: a retriever finds chunks → they're stuffed into a prompt → a language model generates the prose answer. The "G" is the whole point of the name; without a generator writing the answer, it isn't RAG.

Magpie has no G:

1. There is no generator anywhere in the search path. Nothing in Magpie composes a natural-language answer. The closest thing to a model — the cross-encoder reranker — outputs a relevance number per result and reorders the list. It scores; it never writes a sentence.
2. It stops at "here are the ranked hits." A RAG owns the prompt assembly and the model call. Magpie returns the fused, trust-ranked results and hands them back through MCP. What the AI does next — whether it even generates anything — is the AI's job, outside Magpie.
3. Its retriever is more than a RAG's retriever, not less. A textbook RAG retriever is one vector store: embed the query, top-k by cosine, done. Magpie's retrieval is six sources, five modes, trust-weighted fusion, cross-source dedup. It's a far more capable "R" — but it's still only the R.

Plug Magpie into an AI and the pair can form a RAG — Magpie is the R, the AI you bring is the G. But Magpie by itself ships only the R, and a stronger R than usual. It finds and ranks the truth; it never generates the answer.

Deep web search — research breadth without the token bill

The expensive part of "deep research" is reasoning, and the multi-agent approach pays for it N times over — one full LLM context per agent, often millions of tokens for a single question. But reasoning doesn't need to fan out; one capable model already in context can synthesize. Only the searching needs breadth — and searching the web is pure retrieval, zero LLM tokens.

magpie-search deepweb is built on that asymmetry. It fires several sub-queries at the web in parallel, fuses them by trust-weighted RRF + dedup-by-URL into one compact, token-budget-trimmed source set, optionally reads the top pages' text (still token-free), and reports how many independent domains corroborate the result — an agent-free version of the verification a research swarm pays agents to do.

So you get the breadth, page-reading, and corroboration of a multi-agent deep search, but your model only pays for a single synthesis pass over a trimmed result set.

Token cost, measured — one deep question:

That's ~2,000× fewer tokens — about 1/2000th the cost — because the searching and page-reading are pure retrieval (zero model tokens); your model only does the final synthesis pass over the trimmed, corroborated set.

# one question, several angles, read the top pages — all token-free retrieval
magpie-search deepweb "the question" --q "another angle" --q "a third angle" --thorough

The model in your loop then does one synthesis pass over the merged, corroborated set. That's the whole saving: the breadth is free, you pay only for the answer.

Install

pip install magpie-search

Live web search and deepweb need the web extra:

pip install "magpie-search[web]"

Without it, --sources web and deepweb return no results — the web provider is designed to fail soft so it can never break a federated search, so a missing dependency looks exactly like a search that found nothing. As of 1.3.1 it also prints a one-line reason on stderr instead of failing silently.

Or install the latest straight from source:

pip install "git+https://github.com/xfloukiex-lab/magpie-search.git#egg=magpie-search[web]"

Optional — add the local-LLM features (the cross-encoder reranker runs on the base install; the session summarizer needs Ollama):

# 1. Install Ollama (free, runs entirely locally) — https://ollama.com/download
# 2. Pull the model magpie-search uses
ollama pull phi3.5

Python 3.10+ on Windows, macOS, and Linux.

Quickstart

magpie-search index                               # build the index (incremental)
magpie-search search "that retry backoff thing"   # keyword search
magpie-search search --mode hybrid "..."          # keyword + semantic, fused
magpie-search search --mode rerank "..."          # + cross-encoder rerank
magpie-search stats                               # sanity-check the index

Connect it to your AI (MCP)

Magpie speaks the Model Context Protocol, so any MCP-capable agent can call it. Point your client at the bundled server:

// e.g. an MCP client config
{
  "mcpServers": {
    "magpie": { "command": "magpie-search-mcp" }
  }
}

The agent then has search, recent, session, list_sessions, stats, and reindex available — federated, trust-ranked, context-budgeted.

CLI reference

Add --help to any command for full options.

Python API

import magpie_search

results = magpie_search.search("retry backoff", mode="hybrid", k=5)
for h in results["hits"]:
    print(h["trust"], h["source"], h["snippet"])

# LLM features (needs Ollama + phi3.5)
import magpie_search.llm
ranked  = magpie_search.llm.search_rerank(query="retry backoff", k=3, pool=10)
summary = magpie_search.llm.summarize(session_id="abc-123", n_messages=80)

Backup

magpie-search backup copies your transcript tree to a destination of your choice — a local folder (default, zero config), a remote SSH target (NAS / home server), or a remote SSH target with VM boot/suspend. Configure it in ~/.magpie-search/backup.env:

MAGPIE_SEARCH_BACKUP_SSH_HOST=user@nas.local
MAGPIE_SEARCH_BACKUP_SSH_DEST=~/claude-transcripts/

Useful flags: --dry-run, --no-suspend, --show-config. Backup copies; it never deletes originals.

Configuration

Everything is environment-variable driven with sensible defaults.

The summarizer passes through a 6-probe guardrail stack (length, proper-noun-safety, identifier-safety, refusal-drift, semantic-grounding, self-verify); all six must pass for trust: clean. Any failure suppresses the summary and returns trust: degraded — quiet over wrong. Raw messages stay accessible via magpie-search session SESSION-ID.

Privacy

Magpie Search is a local tool. No server, no account, no auto-update, no crash reporter, and no telemetry unless you explicitly opt in (see below). Your transcripts, the index, the audit log, the model cache, and the backups all live on your machine.

Opt-in telemetry. Telemetry is off by default — magpie sends nothing until you run magpie-search telemetry enable (or set MAGPIE_SEARCH_TELEMETRY=1). When on, it sends only anonymous usage: which command ran, search mode, result/hit counts, latency, error class, and your magpie/python/OS versions, tagged with a random install id. It never sends your queries, file paths, results, transcript content, username, or IP — a hard content firewall in telemetry.py drops anything that isn't a number or a short enum token. Disable anytime with magpie-search telemetry disable; check state with magpie-search telemetry status. The only network calls it ever makes are: your local Ollama server (LLM features), your own backup target (only when you run backup), and a one-time model download from Hugging Face on first run. Verify it yourself with tcpdump, Wireshark, or a network-blocked sandbox.

Scheduling

Run magpie-search index (and optionally backup) on a schedule. Ready-made units live in installers/ for systemd (Linux), launchd (macOS), and Task Scheduler (Windows).

Troubleshooting

• "rsync not on PATH" — falls back to scp -r. On Windows, install Git for Windows, which ships rsync.
• Search returns nothing — run magpie-search stats; if last_indexed_at is null, run magpie-search index.
• --sources web or deepweb returns nothing — the web extra is not installed: pip install "magpie-search[web]". Check with python -c "from magpie_search.providers.web import WebProvider; print(WebProvider().health())" — you want ok: True. If it is ok: True and results are still empty, the search engines are refusing your requests, which is common from datacenter or VPS address ranges; try from a different network or set a different engine order via the provider's backends= config.
• Summarizer always degraded — that's the false-positive guard working as designed. Raw transcripts remain available via session SESSION-ID.

About

Magpie Search is built by VektorGeist LLC.

We build local-first tools for people who run their own AI. Magpie is the search core; our agent platform is at vektorgeist.com.

• Website: vektorgeist.com
• Contact: floukie@vektorgeist.com
• Issues & contributions: open an issue or PR on this repository.

License

Licensed under the Apache License 2.0 — see LICENSE. Copyright © 2026 VektorGeist LLC.

"Magpie Search" and the magpie mark are trademarks of VektorGeist LLC. The code is open under Apache-2.0; the brand and name are reserved.