Projekt-Screenshots
Übersicht
XERJ provides an autoindex capability that allows AI agents to efficiently search and retrieve information from local datasets without the token waste associated with tools like grep or sed. By pointing XERJ at a directory, it automatically infers datasets, parses files using tree-sitter for AST-aware code indexing, and makes the data queryable. It implements the Elasticsearch REST API, achieving high conformance (1360/1363 cases), meaning existing Elasticsearch clients, dashboards, and libraries can connect to it directly. It is written in Rust and distributed as a single static binary with no JVM or external dependencies.
Wichtige Funktionen
- Autoindexing capability that infers datasets and requires no schema or pipeline configuration
- Elasticsearch REST API compatibility (passes 1360 of 1363 conformance test cases)
- AST-aware code indexing via tree-sitter for accurate symbol and line number extraction
- Full-text, vector kNN, and hybrid semantic search support
- Single static binary with no JVM or external dependencies
- Supports multiple file formats: CSV, JSON, JSONL, XML, YAML, SQLite, PDF, DOCX, HTML, and common log formats
- 1.72x ingest throughput and 1.61x smaller on-disk footprint compared to Elasticsearch 8.13.4 in benchmarks
Voraussetzungen, Installation und Schnellstart
curl -fsSL https://xerj.org/get | sh
Prebuilt binary (Windows PowerShell):
irm https://xerj.org/get.ps1 | iex
Build from source:
git clone https://github.com/xerj-org/xerj
cd xerj/engine
cargo build --release -p xerj-server
Nutzung
xerj --insecure --data-dir ./data &
Index a folder:
xerj autoindex ~/my-project
Search indexed data:
curl "localhost:9200/ax-*/_search?q=checkout+error"
Structured query with aggregations:
curl localhost:9200/ax-orders/_search -H 'content-type: application/' -d '{
"query": { "range": { "total": { "gte": 100 } } },
"aggs": { "by_status": { "terms": { "field": "status" } } }
}'
Modellkompatibilität und Anwendungsfälle
Not stated in the repository metadata
Lizenz- und Risikohinweise
Apache-2.0
Editorial verification 2026-08-02: repository URL, owner, description, license and repository statistics were reviewed. License metadata: Apache-2.0. README was fetched for the channel draft; re-check repository dependencies, releases and model terms before production use.
Veröffentlichung und Wartung
Not stated in the repository metadata