ltcai 9.0.0 → 9.2.0

package/README.md

@@ -1,9 +1,28 @@
 # Lattice AI
 
-**Lattice AI 9.0.0 is the local-first Digital Brain platform. This release closes the July 8 code-review follow-up, hardens chat/runtime reliability, pays down cleanup debt across runtime utilities and setup detection, and keeps package, desktop, extension, static app, and documentation metadata synchronized to the 9.0.0 line.**
+**Lattice AI 9.2.0 is the local-first Digital Brain platform. Model-Agnostic File Generation makes "create a file" work reliably with any loaded LLM: file requests run through a strict extension-aware prompt, robust extraction, per-type validation, one corrective retry, and a deterministic repair fallback, so even small local models (gemma/qwen class) always produce a structurally valid HTML, JSON, or code file.**
 
 **Lattice AI는 모델이 바뀌어도 내 지식과 맥락을 보존하는 로컬 우선 AI 브레인입니다.**
 
+> The 9.2.0 release hardens file creation end to end: chat file requests are
+> routed to a deterministic direct-write path (including type-only requests
+> like "html 파일 만들어줘"), model replies are treated as untrusted content
+> and cleaned of fences, reasoning blocks, and chat framing, and the agent
+> JSON loop tolerates small-model formatting slips instead of aborting.
+> The 9.1.0 release added request-scoped model routing,
+> workspace-isolated graph identities and frontend caches, fail-closed admin
+> gates, SSRF-safe web capture, private local state permissions, and reproducible
+> release/test isolation. The same release introduced a human-first UI:
+> a visible knowledge journey from conversation or source capture into the
+> living Brain, its real relationship graph, and memory-grounded automation;
+> the empty Brain home now keeps that complete loop inside one viewport, with
+> continuous vital motion and behavior-driven listening, recall, synthesis,
+> and action states. Task navigation and technical detail stay calm and
+> approachable.
+> Telegram access requires an explicit chat allowlist and server session token;
+> public invitation access uses signed, expiring authorization instead of a
+> static cookie; and unknown or unreadable Knowledge Graph scope fails closed.
+
 Your model is the voice you use today. Your Brain is the asset you keep.
 Lattice AI preserves conversations, documents, decisions, project context,
 relationships, and workflows on your computer by default. Cloud models, model
@@ -11,7 +30,9 @@ downloads, update checks, and other external communication happen only after
 explicit consent.
 
 It is not a ChatGPT clone, a model launcher, a graph database, or a note app.
-It is the finished private AI memory layer wrapped in a Living Brain experience — with desktop source capture that uses native folder selection, ToolRegistry readiness, Config-driven DI for automation, and continued AgentRuntime/Tool wiring seams.
+It is a Living Brain: chatting or adding a file, folder, note, or web page grows
+durable memory; the real graph shows how that knowledge connects; and reviewed,
+user-enabled automations can act from the same evidence.
 
 [![PyPI Version](https://img.shields.io/pypi/v/ltcai?label=PyPI)](https://pypi.org/project/ltcai/)
 [![npm Version](https://img.shields.io/npm/v/ltcai?label=npm)](https://www.npmjs.com/package/ltcai)
@@ -43,10 +64,13 @@ You need Lattice AI when:
   session as disposable.
 - Add documents, selected local folders, notes, screenshots, and web pages with
   source-aware memory.
-- See recent memories, older memories, topics, relationships, and the full
-  knowledge graph when you want deeper structure.
-- Create consent-first Brain automation drafts for memory digests, project
-  reviews, and follow-up suggestions before any schedule is enabled.
+- Watch new knowledge enter the Brain and appear in a lightweight, real
+  relationship graph before opening the full graph explorer.
+- See the Brain breathe, pulse, listen, recall, synthesize, and act while the
+  source, graph, composer, and next memory-grounded action remain visible in a
+  single desktop or mobile viewport.
+- Create evidence-linked Brain automation drafts for memory digests, project
+  reviews, and follow-up suggestions, then explicitly enable them when ready.
 - Use a recommended local model without learning model internals first.
 - Keep advanced controls, audit logs, roles, and retention in a separate Admin
   surface.
@@ -58,18 +82,21 @@ You need Lattice AI when:
 2. Create or open a local profile.
 3. Let Lattice explain what this computer can run.
 4. Start with the recommended model as the Brain's voice, or skip and choose later.
-5. Talk to your Brain.
-6. Use the memory rings to move from current context to the full knowledge graph.
-7. Back up, inspect, export, or restore the Brain when you need ownership actions.
+5. Talk to your Brain or add a file, folder, note, or web page.
+6. Watch the source become memory and connect to the visible knowledge graph.
+7. Ask, delegate, review, or explicitly enable a memory-grounded automation.
+8. Back up, inspect, export, or restore the Brain when you need ownership actions.
 
 ## Living Brain Flow
 
 The screenshots below are the latest checked-in visual evidence captures. They
 keep the first-run Brain flow, memory graph, source capture, model library,
 system view, admin console, and review center visible as release gates while
-9.0.0 focuses on code-review closure, chat/runtime reliability,
-AgentRuntime and ToolRegistry maintainability, release artifact correctness,
-and frontend/runtime cleanup.
+9.2.0 makes file generation model-agnostic: chat file requests always end in
+a structurally valid file regardless of which LLM is loaded, backed by
+extraction, validation, corrective retry, and deterministic repair. The
+captures below are the checked-in 9.2.0 visual release evidence for that
+product flow.
 
 ### 1. Wake Brain
 
@@ -81,36 +108,40 @@ confirm owner, check the computer, choose the Brain voice.
 Choose the owner of the Brain. The profile is not a SaaS account by default; it
 is the local identity for the knowledge you keep.
 
-![Login](output/release/v8.7.0/screenshots/01-login.png)
+![Login](output/release/v9.2.0/screenshots/01-login.png)
 
 ### 3. Recommended Models
 
 Start with a short list: safest recommendation, faster model, stronger model.
 Advanced details stay available without overwhelming first-time users.
 
-![Recommended Models](output/release/v8.7.0/screenshots/02-recommended-models.png)
+![Recommended Models](output/release/v9.2.0/screenshots/02-recommended-models.png)
 
 ### 4. Install And Load
 
 Download and load only after consent. Lattice explains model size, local
 execution, and network use before work starts.
 
-![Install and Load](output/release/v8.7.0/screenshots/03-install-load-progress.png)
+![Install and Load](output/release/v9.2.0/screenshots/03-install-load-progress.png)
 
 ### 5. Brain Chat
 
-Talk normally. Useful decisions and context become memory, then appear later as
-topics, relationships, graph structure, and the concentric memory rings around
-the Brain.
+Talk normally or add a source from a single, living canvas. The default home
+fits the complete knowledge lifecycle into one viewport: source controls feed
+the breathing Brain, real nodes and animated relationships stay visible beside
+it, and the composer plus grounded next action remain within reach. Brain motion
+and its visible life signal follow real listening, recall, synthesis, and action
+state. Detailed memory rings, provenance, conversation history, and
+model/runtime proof open as overlays only when requested.
 
-![Brain Brief Home](output/release/v8.7.0/screenshots/04-brain-chat-home.png)
+![One-viewport Living Brain Home](output/release/v9.2.0/screenshots/04-brain-chat-home.png)
 
 ### 6. Review Center
 
 Automation results are staged for review before they become durable decisions.
 Snooze, unsnooze, run now, approve, and dismiss actions stay explicit.
 
-![Review Center](output/release/v8.7.0/screenshots/12-review-center.png)
+![Review Center](output/release/v9.2.0/screenshots/12-review-center.png)
 
 ## Brain Depths
 
@@ -126,10 +157,10 @@ The user travels inward from everyday memory to deeper structure:
 
 Walkthrough:
 
-![v8.7.0 Living Brain walkthrough](output/release/v8.7.0/gifs/v8.7.0-living-brain-walkthrough.gif)
+![v9.2.0 Living Brain walkthrough](output/release/v9.2.0/gifs/v9.2.0-living-brain-walkthrough.gif)
 
 Screenshot index and capture notes:
-[output/release/v8.7.0/SCREENSHOT_INDEX.md](output/release/v8.7.0/SCREENSHOT_INDEX.md)
+[output/release/v9.2.0/SCREENSHOT_INDEX.md](output/release/v9.2.0/SCREENSHOT_INDEX.md)
 
 ## Install
 
@@ -164,7 +195,11 @@ pip install "ltcai[local]"
 - **Product category**: local-first Digital Brain.
 - **Core capability**: private AI memory layer for conversations, documents,
   decisions, relationships, workflows, and project context.
-- **UX metaphor**: Living Brain.
+- **UX metaphor**: a visible source-to-memory-to-graph-to-automation journey
+  centered on the Living Brain, not a generic chat or operations dashboard.
+- **Product navigation**: desktop task navigation and a mobile bottom bar expose
+  Chat, Sources, Memory, and Work; model, workspace, and admin controls live in
+  the secondary menu.
 - **Desktop shell**: Tauri 2 starts a localhost sidecar.
 - **Frontend**: React, TypeScript, Vite, TanStack Query, Zustand, Cytoscape.js,
   React Flow, and generated OpenAPI types.
@@ -210,38 +245,43 @@ See [docs/DEVELOPMENT.md](docs/DEVELOPMENT.md) for developer workflow details.
 
 ## Current Release
 
-The current release is **9.0.0 — Code Review Closure & Runtime Cleanup**:
-
-- July 8 code-review follow-ups are fixed across chat, agent runs, permissions,
-  Telegram integration, runtime audit timestamps, embeddings, and Brain UI copy.
-- Chat/document streaming now preserves terminal SSE events and history/trace
-  persistence when generation fails mid-stream.
-- Runtime audit appends use JSONL while legacy JSON reads remain compatible.
-- Shared JSON/ISO/hash, setup detection, and frontend helper utilities replace
-  duplicated implementations across runtime, Brain Core, setup, and React code.
-- The legacy `server_app` namespace is allowlist-based instead of exposing all
-  app-factory locals.
-- The main `/chat` handler now routes through focused intent handlers and shared
-  response epilogues for lower future regression risk.
-
-Expected artifacts for 9.0.0 release must use exact filenames:
-
-- `dist/ltcai-9.0.0-py3-none-any.whl`
-- `dist/ltcai-9.0.0.tar.gz`
-- `ltcai-9.0.0.tgz`
-- `dist/ltcai-9.0.0.vsix`
-- `src-tauri/target/release/bundle/dmg/Lattice AI_9.0.0_aarch64.dmg`
+The current release is **9.2.0 — Model-Agnostic File Generation**:
+
+- A new `latticeai.core.file_generation` pipeline treats every model reply as
+  untrusted content: extension-aware strict prompting, extraction of the real
+  payload from fences/`<think>` blocks/chat framing, per-type structural
+  validation (complete HTML documents, parseable JSON, CSS rule blocks), one
+  corrective retry that tells the model what was wrong, and a deterministic
+  repair fallback that guarantees a valid file.
+- Type-only requests without an explicit filename ("html 파일 만들어줘",
+  "웹페이지 만들어줘") now resolve to an inferred target and use the
+  deterministic direct-write path instead of the model-driven agent loop.
+- The agent executor tolerates small-model formatting slips: `<think>` blocks
+  and trailing commas are stripped from action JSON, and parse failures feed a
+  corrective format reminder back to the model instead of aborting the run.
+- The executor prompt pins exact file-content rules for `write_file` so
+  agent-created files are complete and fence-free.
+- File writes report whether content was generated cleanly, retried, or
+  auto-repaired in the response payload (`generation` metadata).
+
+Expected artifacts for 9.2.0 release must use exact filenames:
+
+- `dist/ltcai-9.2.0-py3-none-any.whl`
+- `dist/ltcai-9.2.0.tar.gz`
+- `ltcai-9.2.0.tgz`
+- `dist/ltcai-9.2.0.vsix`
+- `src-tauri/target/release/bundle/dmg/Lattice AI_9.2.0_aarch64.dmg`
 
 Do not use wildcard artifact uploads. Package registry publishing remains owner-run.
 
 See [docs/ROADMAP_RECOMMENDATIONS.md](docs/ROADMAP_RECOMMENDATIONS.md) for the
-strategic roadmap slices applied through 9.0.0 and the follow-up tracks.
+strategic roadmap slices applied through 9.2.0 and the follow-up tracks.
 
 ## Known Limitations
 
 - External package registries are owner-published and can lag behind GitHub.
 - PostgreSQL/pgvector is optional scale/migration tooling. SQLite remains the
-  live local Brain store in 9.0.0.
+  live local Brain store in 9.2.0.
 - Docker, model downloads, cloud model calls, Telegram, Brain Network, and update
   checks require explicit user action.
 - Conversation does not fabricate answers when no model is loaded.
@@ -253,6 +293,8 @@ strategic roadmap slices applied through 9.0.0 and the follow-up tracks.
 
 | Version | Theme |
 | --- | --- |
+| 9.2.0 | Model-Agnostic File Generation: chat file requests always produce structurally valid files with any LLM via extraction, per-type validation, corrective retry, deterministic repair, inferred file targets, and a fault-tolerant agent JSON loop |
+| 9.1.0 | Code Review Completion & Fail-Closed Runtime: all July 11 review findings closed across fail-closed security, typed runtime/model/chat boundaries, honest frontend failures and tests, and repository hygiene |
 | 9.0.0 | Code Review Closure & Runtime Cleanup: July 8 code-review follow-ups fixed, chat/runtime reliability improved, duplicated utility surfaces consolidated, runtime audit append paths moved to JSONL, and release metadata/artifacts synchronized |
 | 8.9.0 | Scoped Memory & Tool Policy Hardening: authenticated history/KG reads are workspace-scoped, direct Tool API paths enforce registry policy, local approvals hash tokens at rest, AgentRuntime approval semantics are explicit, and frontend/runtime seams are split |
 | 8.8.0 | Brain Core Extraction & Recall Proof Hardening: internal-only Brain shim layers are removed, AgentRuntime run contracts/retry budgets are tighter, Brain Chat gains conversation controls, and citation recall exposes matched evidence |