AI search engine and MCP server for bots and agents. One Rust binary indexes code, docs, logs, JSONL/CSV and PDFs for BM25 full-text, vector and hybrid search, code search without grep, agent memory and RAG. Elasticsearch-compatible API; Corpus Hub serves hundreds of reference memory corpora.
Rust agent-memoryai-agentsbm25claude-codeclicode-searchcursordatabaseelasticsearchfull-text-searchhybrid-searchjsonllocal-firstmcpmcp-servermodel-context-protocolragrustsearch-enginevector-search
Commits on main
fix(api): delete a one-page _delete_by_query match set without a flush (#2782)
d4a53fa xerj-org committed 20 days ago
fix(autoindex): keep every line when the text chunker hits its byte cap (#2668)
723cad8 Amartya Vikram Singh committed 20 days ago
Add six category pages: hybrid search, BM25 vs vector, Mem0, claude-mem and MemPalace, Claude Code memory, agentic RAG (#2785)
2cc0b16 xerj-org committed 20 days ago
feat(init): teach agents to find an installed skill by task (#2644) (#2748)
cd2f5e9 Jake Elbirt committed 20 days ago
fix(docvalues): reserve read budget for uncached doc_values decodes (Refs #2427) (#2778)
f855d39 xerj-org committed 20 days ago
fix(autoindex): index Vue and Svelte components as code, not XML (#2732) (#2735)
32708f8 Cristian Groza committed 20 days ago
fix(aggs): bound hdr number_of_significant_value_digits; one request hung a search thread (#2651)
6236ecc Amartya Vikram Singh committed 20 days ago
fix(storage): store common doc ids inline in the version map (#2781)
63e1402 xerj-org committed 20 days ago
perf(vector): plan HNSW inserts under a shared lock, in parallel per batch (#2728)
f608853 xerj-org committed 20 days ago
chore(metrics): release downloads 2026-10-31 — 50942 assets, 42360 binaries [skip ci]
a3fa917 xerj-org committed 20 days ago
fix(engine): run at most one background cold stored warm per index (refs #2427) (#2715)
f0409d2 xerj-org committed 20 days ago
fix(engine): stop a search deadlocking on a cold segment's stored-section lock (#2780)
2ec5ac5 xerj-org committed 20 days ago
fix(engine): flush one shard at a time when draining for memory pressure (#2753)
e5675de xerj-org committed 20 days ago
fix(engine): a reopened index runs no catch-up merge until it changes (#2710)
8bd229d xerj-org committed 20 days ago
fix(query): answer simple_query_string fields ["*"] from the inverted index (#2714)
72968c3 xerj-org committed 20 days ago
fix(engine): share one parsed _source between storage and FTS memtables (#2734)
8ca0645 xerj-org committed 20 days ago
fix(governor): stop warming published segments once RSS passes the release level (#2718)
44b9b9a xerj-org committed 20 days ago
fix(vector): chunk HNSW vector storage and account for the graph in the ingest trace (#2724)
66f3a54 xerj-org committed 20 days ago
fix(engine): cap a streak merge write at the memory watermark (#2751)
7d38a98 xerj-org committed 20 days ago
fix(http): trust the OS certificate store alongside bundled roots (#2746) (#2747)
9cd77ce xerj-org committed 20 days ago
feat(autoindex): index a SKILL.md's name and description as fields (#2644) (#2741)
3a323ee Jake Elbirt committed 20 days ago
fix(autoindex): index HTML partials and templates, not junk XML (#2740) (#2743)
89cfb0d Jake Elbirt committed 20 days ago
Rebuild 5 low-traffic answers pages with diagrams, charts and a small interactive (#2742)
a55d55c xerj-org committed 20 days ago
fix(engine): size the publish-time stored warm before decoding it (Refs #2427) (#2737)
06d8140 xerj-org committed 20 days ago
fix(governor): size the bulk analyze reservation by token count (#2730)
f4551c8 xerj-org committed 21 days ago
perf(storage): compact version-map keys and report the map's heap (#2427) (#2726)
a482e3b xerj-org committed 21 days ago