@andespindola/brainlink 1.0.5 → 1.0.7

package/docs/AGENT_USAGE.md

@@ -725,14 +725,17 @@ Available MCP tools:
 - `brainlink_context_packs`
 - `brainlink_search`
 - `brainlink_explain`
+- `brainlink_similar_notes`
 - `brainlink_dedupe`
 - `brainlink_resolve_duplicate`
 - `brainlink_add_note`
+- `brainlink_add_notes`
 - `brainlink_remember`
 - `brainlink_inbox_add`
 - `brainlink_inbox_list`
 - `brainlink_inbox_process`
 - `brainlink_delete_note`
+- `brainlink_delete_notes`
 - `brainlink_add_file`
 - `brainlink_canonicalize_context_links`
 - `brainlink_volatile_add`
@@ -859,7 +862,7 @@ Weak retrieval usually means:
 ## Current Limits
 
 - Search supports FTS, local semantic embeddings and hybrid ranking.
-- Local embeddings are deterministic and provider-free; remote embedding providers are not implemented yet.
+- Local embeddings are deterministic and provider-free. Remote embedding providers (`ollama`, `openai`) are also supported via the `embedding` config block; they fall back to lexical-only search if unreachable at query time.
 - MCP integration is available through the `brainlink-mcp` stdio server.
 - HTTP API is local and unauthenticated.
 - Bucket vaults support S3-compatible `s3://bucket/prefix` URIs and use local cache/index artifacts.

package/docs/ARCHITECTURE.md

@@ -141,11 +141,13 @@ question
   -> selected mode: fts | semantic | hybrid
   -> optional query embedding
   -> optional compressed pack prefilter (token bloom)
-  -> lexical scoring and/or semantic cosine scoring
+  -> field-weighted BM25 lexical scoring and/or semantic cosine scoring
   -> cosine similarity over candidate chunks
+  -> hybrid combines both components on a shared min-max-normalized [0,1] scale
+  -> maximal-marginal-relevance document ordering to reduce near-duplicate context
   -> middle-out context expansion around strongest chunk
   -> ranked chunks with textScore and semanticScore
-  -> token-budget selection
+  -> token-budget selection using the cached per-chunk token count
   -> Markdown context package
   -> if cag/auto, persist derived context pack with current data signature
   -> return cache, strategy and timing metrics
@@ -323,6 +325,5 @@ Parsing, transformation, selection, and formatting are implemented as pure funct
 
 Useful next boundaries:
 
-- remote embedding providers
 - dedicated vector adapter
 - `graph-exporter`

package/docs/QUICKSTART.md

@@ -46,6 +46,8 @@ Optional per-agent retrieval defaults in `brainlink.config.json`:
 
 ```json
 {
+  "autoUpdateCheck": true,
+  "updateCheckIntervalMs": 86400000,
   "agentProfiles": {
     "coding-agent": {
       "defaultSearchMode": "semantic",
@@ -58,6 +60,8 @@ Optional per-agent retrieval defaults in `brainlink.config.json`:
 }
 ```
 
+`autoUpdateCheck` only prints a newer-version notice to `stderr`; it does not install updates automatically. Set `BRAINLINK_NO_UPDATE_CHECK=1` or `"autoUpdateCheck": false` to disable the check.
+
 ## 4) Run Bootstrap Before Work
 
 MCP clients should call `brainlink_bootstrap` first for each vault/agent session, then `brainlink_context`.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@andespindola/brainlink",
-  "version": "1.0.5",
+  "version": "1.0.7",
   "description": "Local-first knowledge memory for agents with Markdown, backlinks, indexing and context retrieval.",
   "type": "module",
   "license": "MIT",
@@ -64,7 +64,7 @@
     "zod": "^4.3.6"
   },
   "overrides": {
-    "hono": "4.12.21",
+    "hono": "4.12.26",
     "qs": "6.15.2"
   },
   "devDependencies": {