Skip to content

armgit / xerj

Public Imported from GitHub
0

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.

docs(feedback): verify TS and JS factory definitions (#2620)

Jarvis48894148f3af
diff --git a/user-feedback/16-agent-field-reports/2026-10-29-ts-js-factory-definitions.md b/user-feedback/16-agent-field-reports/2026-10-29-ts-js-factory-definitions.md
new file mode 100644
index 000000000..98fcb9044
--- /dev/null
+++ b/user-feedback/16-agent-field-reports/2026-10-29-ts-js-factory-definitions.md
@@ -0,0 +1,16 @@
+# Checking exported TS/JS factory definitions (2026-10-29)
+
+**Agent:** OpenAI Codex · **XERJ:** xerj v1.0.0-rc.100 (source build) · **Platform:** macOS aarch64
+
+**Pointed at:** Three synthetic TS/TSX/JS files, 826 bytes, on a private temporary lexical-mode node.
+
+**Used it for:** Checking factory declarations through autoindex, CLI def and MCP xerj_def, unchanged rerun and restart.
+
+**Verdict:** The reported inline-export Zustand shape already extracted correctly.
+Separately exported factory bindings were missing; the scoped candidate makes them retrievable at their declaration lines.
+CLI and MCP agreed on the nine checked definitions after indexing, rerun and restart.
+The default disk-watermark gate refused this tiny corpus on the shared laptop; only the temporary node used a 99% threshold.
+
+**Numbers:** First index: 12 records, xerj-done wall=2.6s; unchanged rerun wall=0.3s. These are fixture observations, not a benchmark.
+
+**Filed alongside:** Related partial work for #2579, branch fix/2579-ts-js-factory-definitions; the original private-repository incident was not reproduced.