docs: add missing feature docs for Matrix E2EE thumbnails, LINE media, and CJK memory

- Matrix: note encrypted thumbnail behavior in E2EE rooms (#54711)
- LINE: add outbound media section for image/video/audio sends (#45826)
- Memory: document CJK trigram tokenization and chunk sizing

docs/channels/line.md

@@ -184,6 +184,16 @@ The LINE plugin also ships a `/card` command for Flex message presets:
 /card info "Welcome" "Thanks for joining!"
 ```
 
+## Outbound media
+
+The LINE plugin supports sending images, videos, and audio files through the agent message tool. Media is sent via the LINE-specific delivery path with appropriate preview and tracking handling:
+
+- **Images**: sent as LINE image messages with automatic preview generation.
+- **Videos**: sent with explicit preview and content-type handling.
+- **Audio**: sent as LINE audio messages.
+
+Generic media sends fall back to the existing image-only route when a LINE-specific path is not available.
+
 ## Troubleshooting
 
 - **Webhook verification fails:** ensure the webhook URL is HTTPS and the

docs/channels/matrix.md

@@ -189,6 +189,8 @@ done:
 
 ## Encryption and verification
 
+In encrypted (E2EE) rooms, outbound image events use `thumbnail_file` so image previews are encrypted alongside the full attachment. Unencrypted rooms still use plain `thumbnail_url`. No configuration is needed — the plugin detects E2EE state automatically.
+
 ### Bot to bot rooms
 
 By default, Matrix messages from other configured OpenClaw Matrix accounts are ignored.

docs/reference/memory-config.md

@@ -422,6 +422,7 @@ Notes:
 - `vectorWeight` + `textWeight` is normalized to 1.0 in config resolution, so weights behave as percentages.
 - If embeddings are unavailable (or the provider returns a zero-vector), we still run BM25 and return keyword matches.
 - If FTS5 can't be created, we keep vector-only search (no hard failure).
+- **CJK support**: FTS5 uses configurable trigram tokenization with a short-substring fallback so Chinese, Japanese, and Korean text is searchable without breaking mixed-length queries. CJK-heavy text is also weighted correctly during chunk size estimation, and surrogate-pair characters are preserved during fine splits.
 
 This isn't "IR-theory perfect", but it's simple, fast, and tends to improve recall/precision on real notes.
 If we want to get fancier later, common next steps are Reciprocal Rank Fusion (RRF) or score normalization