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
docs(changelog): three audit corrections — gate calibration, 948 residual marker, unit slip
xerj-org4a35e54f9bb6
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex faee56439..5a94d8872 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -22,10 +22,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0reconcile) drops idle CPU to 0.15–0.18 % and wakeups from ~46/s to ~26/s onthe same corpus, with the WAL walk moved to the blocking pool. A scrape isthe only moment these gauges are observable, so nothing is lost between- scrapes. The fixture and its coarse CI gate (`idle-budget` job) keep the- whole budget honest: CPU < 0.5 % of one core, O(1) wakeups/s, ≤ 0.2 MB RSS- per idle index, boot-to-green with zero WAL replay on a cleanly-flushed- corpus. (PR [#1020](https://github.com/xerj-org/xerj/pull/1020).)+ scrapes. The fixture and its coarse CI gate (`idle-budget` job, calibrated+ to 256 kB/idx on the runner's host shape) keep the budget honest: CPU < 0.5 %+ of one core, O(1) wakeups/s, boot-to-green with zero WAL replay on a+ cleanly-flushed corpus. The 0.2 MB/index product line itself was still at or+ above the line on 4-vCPU hosts at this point (204.9–209.9 kB measured,+ N-slope 206.5 kB/idx) — that is+ [#1024](https://github.com/xerj-org/xerj/issues/1024), met only after the+ lazy seen-set work below (~63–69 kB, >3× margin). (PR+ [#1020](https://github.com/xerj-org/xerj/pull/1020).)- **The flush publication bracket no longer spans the drain**([#1015](https://github.com/xerj-org/xerj/issues/1015)): flush drains nowfreeze and the bracket wraps only the publish, so the ms-scale bracket@@ -253,7 +258,8 @@ locally. Everything else in this cut is a fix with a reproduction behind it.- **Ingest memory: a 1 GB mailbox no longer aborts under the default 16 GiBcap, and at-rest settled memory halved**([#948](https://github.com/xerj-org/xerj/issues/948), PR #1002). A 1 GB- mailbox peaked at 25.4 GiB VmHWM under the 16 GiB cap (aborted: 510/2,530+ mailbox peaked at 25,356.9 MiB (24.8 GiB) VmHWM under the 16 GiB cap+ (aborted: 510/2,530needles missing, 0 edges; 68.5 GiB uncapped). The ingest-memory ledger foundfour causes: `memtable_shards` retained one parsed `Arc<Value>` perexplicit-id write forever (flush drained the FTS memtable, never the storage@@ -268,7 +274,10 @@ locally. Everything else in this cut is a fix with a reproduction behind it.972.8 -> 481-488 MiB (variant A, 3 runs) and 883.0 -> 280.8 MiB (variant C,100k x 100 B; retention ~677 -> ~108 MB). Merge batching follows thequickwit/tantivy approach (streaming_writer.rs:378, log_merge_policy.rs:22 —- no code copied).+ no code copied). Residual, separately tracked: three per-segment caches+ (`stored_value_cache`, `dv_cache`, `id_pos_cache`) remain unbounded until a+ merge retires their segment —+ [#1032](https://github.com/xerj-org/xerj/issues/1032).- **scalar8 keeps its promise: 2.3x faster kNN, and `_score` no longer dependson the filter**([#392](https://github.com/xerj-org/xerj/issues/392), PR #999). scalar8