📜 membank: add specification document and sline updates

Added membank feature specification document outlining the evolution from libsource to membank system:

- membank/spec.md - New feature specification for enhanced RAG system
- Updated sline-db.json with current emoji rotation state

The membank system represents the next evolution of our libsource RAG implementation,
providing a foundation for multiple entity types (libsources, docs, etc.) all served
through the unified 1408 API.

🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>

Author: dvakatsiienko

.clauderc/sline/sline-db.json

@@ -1 +1 @@
-{"current_index":1,"last_update_time":1757437359}
+{"current_index":3,"last_update_time":1757455463}

membank/spec.md

@@ -0,0 +1,55 @@
+# new «membank» feature
+
+let's refine «libsource» feature. new codename for it — «membank» (root name). example use case:
+
+- «hey claude, search membank about vite custom plugin implementation usage»
+- claude: searches 1408 port for relevant vite info
+
+key entities:
+
+- membank — root feature name
+- libsource — sub-entity, means a library source extracted via gitingest to a .txt file, and then
+  converted in rag augmented shape into db
+- since any kind of data may be rag-enhanced we likley extend our db with more entities.
+    - for example (do not remember this information, it is explanation for the future):
+      libsource-react is a gitingested react github repositroy source. in future we may download
+      react.dev docs and place them alongside libsource-react as docs-react. docs-react will be
+      another type of rag-augmented entity.
+
+location: this directory is a new place to colocate libsources, rag, db and all scripts. previously
+we were relying on symlinks to read .txt libsources. but since we now have a server 1408 that serves
+rag-augmented data, we no longer need filesystem for data access.
+
+#### new structure
+
+- membank
+    - libsource (dir)
+        - core → contains all py crud libsource scripts
+        - /Users/dima/.dotfiles/.clauderc/.membank/libsource-scripts/config/libsource-config.json →
+          moves here and named as .libsource-config.json (start with dot for easier lookup)
+        - list of libsources libsource-\*.txt
+    - cli (dir, contains search.py, server.py)
+    - membank.db (file)
+    - rag (stays as it is)
+    - scripts (stays as it is)
+
+#### key todos:
+
+- move /Users/dima/.dotfiles/.clauderc/.membank/libsource and
+  /Users/dima/.dotfiles/.clauderc/.membank/libsource-scripts in this directory and reconstruct
+  according to «new structure»
+- since all paths are changed, update all path refs
+- delete all libsources.db in this project
+- create new membank.db (empty)
+- test full crud of currently registered libsources — this will uncover bugs after reconstruction
+  and path updates — fix them all
+- after reconstruction and testing full functionality and fixing all issues:
+    - delete all current «libsource» information from ~/.dotfiles/claude.md
+    - create ~/.dotfiles/membank/CLAUDE.md with full details about «membank» feature
+    - update ~/.dotfiles/claude.md again but (important) only mention in high-level manner that
+      «membank» feature source is located in this repository, briefly explain where to look for key
+      entities, and tell that core info located in nested claude.md (probably via reference
+      @membank/CLAUDE.md)
+    - the idea is to modularize memory. ~/.dotfiles/claude.md should contain only a pointer to
+      membank (not full info), but full info is in nested claude.md
+- update package.json refs