xerj-org
XERJ is an AI-native local search engine distributed as a single Rust binary. It indexes code, documents, logs and other data into typed, queryable indices so AI agents can retrieve the relevant definition or passage instead of rereading whole files and spending context-window tokens.
At a Glance
- Open-source projects and individual developers
- Coding-agent and AI application teams
- Mid-size enterprises needing high-availability local search
- Regulated and global enterprises needing multi-region or air-gapped deployment
- +2 more
AI Tools by xerj-org
(1)
XERJ
Rust Search Engine for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about xerj-org
Latest News
The exploit hub, tested on three fresh CVEs: XERJ evaluated retrieval over a 1.09-million-document exploit corpus and published the losses as well as wins.
Six eval scars from a production agent-memory system: a guest evaluation by Chunxiao Wang, with XERJ documenting the gap in its own agent-trajectory corpus.
The Corpus Hub: pinned, licensed, signed and contributed by pull request, with a 2.7x output-token reduction measured for reference coding.
The rc.81 release gates: three gates passed, three failed and one split, with public measurements for ask, mapping, calibration and judge behavior.
Products & Services
Apache-2.0 single Rust binary providing local full-text search, vector/kNN search, hybrid retrieval, typed storage and agent memory over an Elasticsearch-8.x-compatible REST API.
Zero-configuration folder indexer that content-sniffs and streams code, JSON/JSONL, CSV, SQL, SQLite, PDF, DOCX, HTML, XML, YAML, logs, mailboxes and other formats into typed indices, with resumable and idempotent ingestion.
Bundled console and browser playground for querying indices and viewing live or seeded dashboards for retrieval, vectors, agent memory, logs, anomalies and ingest.
Registry and signed-pack workflow for reproducible reference corpora: manifests pin source repositories and commit SHAs, while corpus packs are built, signed and indexed locally.
Market Position
XERJ positions itself as a local, agent-oriented alternative to assembling Elasticsearch, a vector database, an embedding service and a separate memory store. Its differentiators are one static binary, zero-config mixed-format indexing, AST-aware code retrieval, ES-compatible clients, offline operation and a single hybrid-search/memory surface. The site's comparison set includes Elasticsearch, Meilisearch, Typesense, Pagefind, ripgrep/ripgrep-all, ast-grep, DocFetcher, HoudahSpot, Recoll, Khoj, LocalSynapse, Obsidian Omnisearch, standalone vector databases and web-agent search. XERJ's own benchmarks also publish cases where Elasticsearch or hosted Jev performs better.
Founding Story
The project was started around the problem that coding agents often invent APIs or enter fix loops when they rely on memory or grep and then have to read whole source files into context. Its initial vision was a single, local, zero-configuration engine that combines full-text search, vector retrieval, agent memory and an Elasticsearch-compatible API without a JVM, separate vector database or external embedding key.
Business Model
Revenue Model
Self-hosted software with a free Apache-2.0 Community edition and paid per-node Professional licensing; Enterprise is custom-priced. The site says there are no per-document, per-vector or per-query fees and customers own their compute.
Pricing Tiers
Full feature set for a single node; community support through GitHub and forum; no SLA.
Three-node HA cluster, synchronous replication, business-hours support, 24-hour response, 99.95% availability SLA, Helm chart and Terraform module.
Multi-region and regulated-enterprise deployment, air-gapped operation, 24x7 support with one-hour response, 99.99% availability SLA, dedicated CSM/solutions architect, HIPAA BAA and custom compliance.
Target Markets
- Open-source projects and individual developers
- Coding-agent and AI application teams
- Mid-size enterprises needing high-availability local search
- Regulated and global enterprises needing multi-region or air-gapped deployment
- Financial services, healthcare, public sector and retail
- Security, observability, SIEM and data/analytics teams
- Reference coding: retrieve mechanisms and definitions from similar open-source repositories before writing code
- Codebase question answering and RAG
- Security audits and code security analysis
- Full-text and semantic search across local document folders
- Log analytics, SIEM and incident investigation
- Operational intelligence and observability across logs, metrics and traces