ltcai 1.2.0 → 1.4.0

package/README.md

@@ -36,6 +36,24 @@ Automatic knowledge graph
 Graph-aware chat, snapshots, memory, agents, workflows, skills, and timeline
 ```
 
+### New in 1.4.0: Server App Final Decomposition
+
+- **server_app.py final decomposition** — the app shell is now FastAPI assembly,
+  lifespan, middleware, static mounting, and router wiring only
+  (~5,381 → 1,303 lines)
+- **Chat / model / tools extraction** — chat/history/agent, model runtime and
+  provider helpers, tools, local files, computer-use, permissions, upload,
+  garden/setup/static UI, MCP, and KG glue now live in API routers and services
+- **AppContext and dependency cleanup** — routers receive explicit dependencies
+  and do not import the FastAPI app; service modules own runtime and dispatch
+  business logic
+- **Safety validation suite** — route compatibility, import/startup, streaming,
+  model endpoint, tools/local/CU, release-artifact, and documentation stale
+  checks guard the split
+- **Compatibility preserved** — all public API paths, request/response schemas,
+  `server:app`, CLI, UI, Knowledge Graph, Admin/Security, Workspace OS, and
+  VS Code expectations remain unchanged
+
 ### New in 1.2.0: Server App Modularization
 
 - **server_app.py modularized** — Workspace/Organization and health/engine
@@ -315,16 +333,18 @@ Supported routes include OpenAI-compatible APIs, OpenRouter, Groq, Together, xAI
 
 ## Current release
 
-**0.6.0** completes the runtime / registry / config extraction sprint:
+**1.4.0** completes the Server App Final Decomposition release:
 
-- `server.py` is now a thin compatibility entrypoint; FastAPI app assembly lives
-  in `latticeai.server_app`
-- tool dispatch, governance, permission views, MCP descriptions, and prompt
-  catalog metadata are centralized in `ToolRegistry`
-- agent role prompts are split into `latticeai.core.agent_prompts`, while
-  `AgentRuntime` remains the injected state-machine core
+- `server.py` remains the thin compatibility entrypoint and
+  `latticeai/server_app.py` is now a compact app assembly shell
+- chat/history/agent, model runtime/provider helpers, tools/local/CU/
+  permissions/upload, garden/setup/static pages, MCP, and KG router wiring are
+  extracted into `latticeai/api/*` and `latticeai/services/*`
+- route compatibility, streaming, model endpoint, tools/local/CU,
+  import/startup, build, packaging, and documentation stale-reference checks
+  are part of the release validation
 - Python package, npm package, VS Code extension, FastAPI app, and `/health`
-  version metadata are aligned at `0.6.0`
+  version metadata are aligned at `1.4.0`
 
 See the full [changelog](docs/CHANGELOG.md).
 

package/docs/CHANGELOG.md

@@ -1,5 +1,101 @@
 # Changelog
 
+## [1.4.0] - 2026-05-31
+
+> Server App Final Decomposition — chat, model runtime, tools/local/CU,
+> permissions/upload, garden/setup/static, MCP, and KG glue extracted while
+> preserving the public route contract.
+
+### Added
+
+- **Final decomposition guard** —
+  `tests/unit/test_server_app_v14_decomposition.py` asserts
+  `latticeai/server_app.py` stays under the 1,500-line target, new routers and
+  services import independently, and version metadata is aligned.
+- **New routers** — `latticeai/api/chat.py`, `latticeai/api/tools.py`,
+  `latticeai/api/computer_use.py`, `latticeai/api/local_files.py`,
+  `latticeai/api/permissions.py`, `latticeai/api/garden.py`,
+  `latticeai/api/setup.py`, `latticeai/api/static_routes.py`, plus
+  `latticeai/api/deps.py`.
+- **New service seams** — `latticeai/services/model_runtime.py`,
+  `latticeai/services/tool_dispatch.py`, `latticeai/services/upload_service.py`,
+  and
+  `latticeai/services/app_context.py`.
+
+### Changed
+
+- **server_app.py final decomposition** — reduced from 5,381 lines to 1,303
+  lines. The file now owns FastAPI construction, lifespan, middleware, static
+  mount, router wiring, and compatibility globals only.
+- **Chat/history/agent extracted** — `/chat`, `/history*`, `/agent*`, streaming
+  generator, document-generation session handling, Graph RAG trace recording,
+  and AgentRuntime wiring moved to `latticeai/api/chat.py` with behavior and
+  SSE chunk format preserved.
+- **Model runtime/provider extracted** — provider catalogs, engine aliases,
+  install/download/pull/load/unload helpers, prepare-model streaming,
+  compatibility smoke tests, runtime feature payloads, and cloud verification
+  moved to `latticeai/services/model_runtime.py`.
+- **Tools/local/CU/permissions/upload extracted** — `/tools/*` moved to
+  `latticeai/api/tools.py`, `/local/*` and KG/local-knowledge router glue moved
+  to `latticeai/api/local_files.py`, `/cu/*` moved to
+  `latticeai/api/computer_use.py`, `/permissions/*` moved to
+  `latticeai/api/permissions.py`, and `/upload/document` now delegates to
+  `latticeai/services/upload_service.py`.
+- **Garden/setup/static routes extracted** — `/garden*`, `/setup*`,
+  `/permissions/open/*`, `/`, `/account`, `/chat`, `/admin`, `/status`,
+  `/manifest.json`, `/sw.js`, and `/local/sysinfo` moved to dedicated routers.
+- **Docs and release metadata aligned** — README current release conflict fixed,
+  SECURITY supported versions updated, package metadata bumped to `1.4.0`, and
+  publish docs avoid unsafe `dist/*` upload commands.
+
+### Validation
+
+- Route compatibility snapshot, import/startup checks, chat streaming contract,
+  model endpoint presence, MCP/KG presence, v1.4 line-count/import/version
+  guard, unit/integration suites, Python build, VSIX package, npm pack, twine
+  check, and release artifact validation all pass for `1.4.0`.
+
+## [1.3.0] - 2026-05-31
+
+> Server app decomposition (phase 3) — safety-net suite first, then model & MCP router extraction.
+
+### Added
+
+- **Route-compatibility safety net** — `tests/unit/test_route_compatibility.py`
+  freezes the full public route surface (209 paths) plus import/startup,
+  streaming-contract, model/engine, and MCP/KG presence checks. Any dropped or
+  renamed endpoint, broken import, or removed `StreamingResponse` now fails the
+  suite immediately. This was built **before** moving code, per the decomposition
+  plan.
+- **Model / engine router** — `latticeai/api/models.py` (`create_models_router`)
+  now owns `/models*`, `/engines*` (install / verify-cloud / pull-model /
+  prepare-model[/stream]) and `/setup/set-api-key`. Heavy provider/runtime
+  helpers remain injected from server_app (no import cycle, no new import-time
+  side effects).
+- **MCP / skills / plugins router** — `latticeai/api/mcp.py` (`create_mcp_router`)
+  now owns `/mcp/*`, `/skills/*`, `/plugins/directory*`, and `/mcp/call`.
+  Registry/tool symbols are imported directly from `mcp_registry` / `tools` /
+  `tool_registry`; server_app-defined helpers are injected.
+
+### Changed
+
+- **server_app.py decomposition** — reduced from ~5,948 to ~5,382 lines by
+  extracting the model/engine and MCP/skills/plugins clusters (and their
+  request models) into the routers above. All API paths, request/response
+  schemas, the `server:app` import path, CLI, UI, KG/Admin/Security routers, and
+  VS Code integration are unchanged (asserted by the route snapshot test).
+- Release metadata aligned to `1.3.0`; `/health` reports `1.3.0`.
+
+### Notes
+
+- The chat/streaming cluster, the `/tools/*` · `/cu/*` · `/local/*` ·
+  `/upload` · `/permissions` clusters, and the ~1,700-line model/engine
+  *provider helper* block remain in server_app and are scheduled for the next
+  decomposition pass (the safety net now de-risks those moves). server_app.py is
+  not yet under the 2,000-line target.
+- CI hardening from 1.0.1/1.1.0 retained (VSIX compile guard, Node.js 24,
+  version-scoped artifact validation — no `dist/*` glob).
+
 ## [1.2.0] - 2026-05-31
 
 > Server app modularization (routers + service layer) and workspace/org guardrail hardening.

package/latticeai/__init__.py

@@ -1,3 +1,3 @@
 """Lattice AI - modular server package."""
 
-__version__ = "1.2.0"
+__version__ = "1.4.0"