ltcai 7.9.0 → 8.1.0

package/README.md

@@ -1,6 +1,6 @@
 # Lattice AI
 
-**Lattice AI 7.9 is the local-first Digital Brain product with cleaner runtime boundaries. It keeps your knowledge durable across any AI model, with Brain Chat Home ready for immediate use and agent execution hardened behind explicit runtime ports.**
+**Lattice AI 8.0 is the local-first Digital Brain platform with hardened runtime architecture. It keeps your knowledge durable across any AI model, with AgentRuntime, ToolRegistry, central Config, and Knowledge Graph stability tracked by explicit release contracts.**
 
 **Lattice AI는 모델이 바뀌어도 내 지식과 맥락을 보존하는 로컬 우선 AI 브레인입니다.**
 
@@ -11,7 +11,7 @@ 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 — now with a clearer 7.9 agent-runtime boundary.
+It is the finished private AI memory layer wrapped in a Living Brain experience — now with the 8.0 runtime architecture contract behind it.
 
 [![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)
@@ -64,9 +64,9 @@ You need Lattice AI when:
 
 ## Living Brain Flow
 
-The screenshots below are release evidence captures. The 7.9 line keeps the
-same first-run flow and rebuilds the Brain Chat Home layout so chat and the
-composer are the first visible action.
+The screenshots below are release evidence captures. The 8.1 line keeps the
+first-run flow and refreshes Brain Home so the living Brain, recent memory,
+next action, and composer are understandable in one glance.
 
 ### 1. Wake Brain
 
@@ -78,43 +78,36 @@ 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/v7.8.0/screenshots/01-login.png)
+![Login](output/release/v8.1.0/screenshots/01-login.png)
 
-### 3. Environment Analysis
-
-See what kind of local AI experience this computer can support before choosing a
-model.
-
-![Environment Analysis](output/release/v7.8.0/screenshots/02-environment-analysis.png)
-
-### 4. Recommended Models
+### 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/v7.8.0/screenshots/03-recommended-models.png)
+![Recommended Models](output/release/v8.1.0/screenshots/02-recommended-models.png)
 
-### 5. Install And Load
+### 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/v7.8.0/screenshots/04-install-load-progress.png)
+![Install and Load](output/release/v8.1.0/screenshots/03-install-load-progress.png)
 
-### 6. Brain Chat
+### 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.
 
-![Brain Chat Home](output/release/v7.8.0/screenshots/05-brain-chat-home.png)
+![Brain Chat Home](output/release/v8.1.0/screenshots/04-brain-chat-home.png)
 
-### 7. Review Center
+### 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/v7.8.0/screenshots/13-review-center.png)
+![Review Center](output/release/v8.1.0/screenshots/12-review-center.png)
 
 ## Brain Depths
 
@@ -130,10 +123,10 @@ The user travels inward from everyday memory to deeper structure:
 
 Walkthrough:
 
-![v7.8.0 Living Brain walkthrough](output/release/v7.8.0/gifs/v7.8.0-living-brain-walkthrough.gif)
+![v8.1.0 Living Brain walkthrough](output/release/v8.1.0/gifs/v8.1.0-living-brain-walkthrough.gif)
 
 Screenshot index and capture notes:
-[output/release/v7.8.0/SCREENSHOT_INDEX.md](output/release/v7.8.0/SCREENSHOT_INDEX.md)
+[output/release/v8.1.0/SCREENSHOT_INDEX.md](output/release/v8.1.0/SCREENSHOT_INDEX.md)
 
 ## Install
 
@@ -210,36 +203,27 @@ See [docs/DEVELOPMENT.md](docs/DEVELOPMENT.md) for developer workflow details.
 
 ## Current Release
 
-The current release is **7.9.0 — Agent Runtime Boundary Hardening**:
-
-- The single-agent state machine is now named `SingleAgentRuntime`, while the
-  historical `AgentRuntime` import remains as a compatibility alias.
-- The product AgentRuntime facade in `lattice_brain.runtime.agent_runtime`
-  remains the unambiguous `/agents` execution and observability boundary.
-- Agent rollback is now an injected runtime port wired through
-  `ToolDispatchService`, so the core state machine no longer shells out directly.
-- Both runtime surfaces expose a `runtime-boundary/v1` descriptor so tests and
-  API clients can identify the owning execution boundary without relying on
-  ambiguous class names.
-- A shared `RuntimeBoundaryProtocol` now fixes the common inspection surface
-  while letting each runtime keep its own execution methods.
-- Product readiness is machine-checkable through `product_readiness()`, which
-  composes the architecture contract with exact artifact/docs/workflow evidence.
-- CI and release workflows now run frontend lint/typecheck/build gates plus the
-  7.9 product readiness score before a tag can be treated as release-safe.
-
-Expected artifacts for 7.9.0 release must use exact filenames:
-
-- `dist/ltcai-7.9.0-py3-none-any.whl`
-- `dist/ltcai-7.9.0.tar.gz`
-- `ltcai-7.9.0.tgz`
-- `dist/ltcai-7.9.0.vsix`
-- `src-tauri/target/release/bundle/dmg/Lattice AI_7.9.0_aarch64.dmg`
+The current release is **8.1.0 — Intuitive Brain Home**:
+
+- Brain Home now opens with the living Brain directly above the conversation composer, so the primary action is visible before users read any utility panels.
+- The previous dashboard-style growth strip is replaced by a product narrative: Brain state, what was just remembered, the strongest connected topic, and the next best action.
+- Primary actions are explicit and click-tested: talk to Brain, add sources, or open the memory graph.
+- Mobile and narrow layouts keep the Brain and composer in the first viewport, avoiding the previous clipped/overcrowded first screen.
+- The 8.0 runtime architecture contract remains active: AgentRuntime, ToolRegistry, Config, server decomposition, and KG hardening stay machine-checkable through readiness gates.
+- CI and release workflows continue to run frontend lint/typecheck/build gates plus product readiness before a tag can be treated as release-safe.
+
+Expected artifacts for 8.1.0 release must use exact filenames:
+
+- `dist/ltcai-8.1.0-py3-none-any.whl`
+- `dist/ltcai-8.1.0.tar.gz`
+- `ltcai-8.1.0.tgz`
+- `dist/ltcai-8.1.0.vsix`
+- `src-tauri/target/release/bundle/dmg/Lattice AI_8.1.0_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 7.9.0 and the follow-up tracks.
+strategic roadmap slices applied through 8.1.0 and the follow-up tracks.
 
 ## Known Limitations
 
@@ -256,6 +240,8 @@ strategic roadmap slices applied through 7.9.0 and the follow-up tracks.
 
 | Version | Theme |
 | --- | --- |
+| 8.1.0 | Intuitive Brain Home: living Brain, recent memory, connected topic, next action, and composer are visible in one product-first screen with refreshed 8.1.0 evidence and artifacts |
+| 8.0.0 | Runtime Architecture Contract: AgentRuntime, ToolRegistry, central Config, server decomposition, and KG hardening are captured as machine-checkable release boundaries with exact 8.0.0 artifacts |
 | 7.9.0 | Agent Runtime Boundary Hardening: explicit `SingleAgentRuntime`, compatibility alias preservation, injected rollback port, and release/readiness docs aligned to the product AgentRuntime facade |
 | 7.8.0 | Brain Chat Home UX Simplification: chat-first first viewport, visible workspace navigation, collapsed source/status utilities, hidden default depth controls, and removal of obsolete Brain UX components |
 | 7.7.0 | Complete Product Polish: command-center Brain Home, repaired Review Center evidence, exact 7.7 docs/artifacts, product readiness gate, and stronger CI/release checks |

package/docs/CHANGELOG.md

@@ -6,6 +6,41 @@ existed at that release.
 
 ## [Unreleased]
 
+## [8.1.0] - 2026-06-27
+
+### Changed
+- Rebuilt Brain Home around an intuitive first screen with LivingBrain, recent
+  memory, connected topic, next-best action, and the composer visible together.
+- Replaced dashboard-style Brain growth metrics with narrative, product-facing
+  copy and focused primary actions.
+- Tightened mobile and narrow viewport behavior so the Brain and composer remain
+  visible without horizontal overflow.
+- Refreshed release screenshots, walkthrough GIF/WebM, static app assets, and
+  exact 8.1.0 artifact metadata.
+- Synchronized Python, npm, VS Code extension, Tauri, workspace, readiness, and
+  current-release documentation versions to 8.1.0.
+
+## [8.0.0] - 2026-06-24
+
+### Changed
+- Added `lattice-architecture-contract/v1` to make the AgentRuntime,
+  ToolRegistry, Config, server decomposition, and Knowledge Graph stabilization
+  boundaries explicit and testable for the major architecture line.
+- Added `tool-registry-contract/v1` to the ToolRegistry manifest, including
+  dispatch, policy, and permission ownership.
+- Updated architecture and product readiness targets to 8.0.0.
+- Synchronized package/runtime/static/Tauri metadata to 8.0.0.
+- Updated current-release docs and exact artifact names to 8.0.0 while
+  preserving historical 7.x entries.
+
+### Fixed
+- Made logical Knowledge Graph `replace` imports transactional so malformed
+  imports roll back without clearing the existing graph.
+- Locked v2 read-equivalence coverage for `list_documents`, `get_node`,
+  `relationship_search`, and `traverse`.
+- Preserved colliding legacy edge labels during logical import/backfill while
+  keeping native write-door synonym dedupe canonical.
+
 ## [7.9.0] - 2026-06-23
 
 ### Changed

package/docs/V3_FRONTEND.md

@@ -1,10 +1,15 @@
 # Lattice AI v3 — Frontend Product Shell
 
+> Historical v3 shell notes. The current product shell is the React/Vite app
+> built from `frontend/` into `static/app/` with `npm run build:assets`; current
+> linting uses `scripts/lint_frontend.mjs`.
+
 > A token-native single-page workspace shell for Lattice AI v3. It ships as the
 > primary product experience and calls the real v3 retrieval/chat APIs while
 > rendering clear unavailable states for local services that are not running.
 
 Entry point: **`/app`** (served by `latticeai/api/static_routes.py` →
+the current `static/app/index.html` bundle; the historical v3 shell used
 `static/v3/index.html`). Login, auto-login after registration, SSO callback, and
 the PWA manifest land on `/app`. Legacy multi-page static screens have been
 deleted; compatibility routes (`/workspace`, `/chat`, `/graph`, `/admin`, …)
@@ -27,7 +32,8 @@ Deep-linking into an `admin/*` route auto-promotes the shell to Admin mode.
 
 ## Information architecture
 
-One declarative table (`static/v3/js/core/routes.js`) drives the nav rail, the
+One declarative table (historically `static/v3/js/core/routes.js`, now
+represented by the React routes under `frontend/src/`) drives the nav rail, the
 command palette, the router, breadcrumbs, and lazy view loading.
 
 ```
@@ -50,8 +56,8 @@ Administration Users · Permissions · Audit Logs · Security · Policies · Pri
 Token-native, layered on the existing **single color source**
 (`static/css/tokens.css`, `data-lt-theme` light/dark). No legacy override layers
 (`responsive.css` / `workspace.css` / `platform.css`) are loaded by v3.
-Runtime assets are loaded through `static/v3/asset-manifest.json`; the build
-step writes hashed CSS/JS siblings and `/app` reads the manifest instead of
+Runtime assets are loaded through `static/app/asset-manifest.json`; the build
+step writes hashed CSS/JS assets and `/app` reads the manifest instead of
 using manual `?v=` query strings.
 
 | File | Responsibility |
@@ -133,8 +139,8 @@ focus rings, skip link, keyboard command palette (⌘K / Ctrl-K).
 
 ## Validation
 
-- `npm run lint` (extended to cover `static/v3/**` via `scripts/lint_v3.mjs`).
-- `npm run build:assets` (writes hashed v3 assets and `asset-manifest.json`).
+- `npm run lint` (covers the current frontend through `scripts/lint_frontend.mjs`).
+- `npm run build:assets` (writes hashed `static/app` assets and `asset-manifest.json`).
 - `npm run test:visual` (`tests/visual/v3.spec.js` against the mock server, which
   serves `/app` and mocks the future API surfaces).
 - Browser-rendered smoke checks of every route in light and dark themes.

package/lattice_brain/__init__.py

@@ -26,7 +26,7 @@ from .storage import (
     storage_from_env,
 )
 
-__version__ = "7.9.0"
+__version__ = "8.1.0"
 
 __all__ = [
     "AgentRuntime",

package/lattice_brain/embeddings.py

@@ -15,6 +15,10 @@ DEFAULT_EMBEDDING_DIM = int(os.getenv("LATTICEAI_VECTOR_DIM", "384"))
 EMBEDDING_MODEL_ID = f"lattice-local-hash-v1:{DEFAULT_EMBEDDING_DIM}"
 
 
+def embedding_model_id(dim: int) -> str:
+    return f"lattice-local-hash-v1:{int(dim)}"
+
+
 def _tokenize(text: str) -> List[str]:
     raw = str(text or "").lower()
     tokens = re.findall(r"[a-z0-9][a-z0-9_.:/+-]{1,}|[가-힣]{2,}", raw)
@@ -49,6 +53,10 @@ class LocalEmbeddingModel:
     dim: int = DEFAULT_EMBEDDING_DIM
     model_id: str = EMBEDDING_MODEL_ID
 
+    def __post_init__(self) -> None:
+        if self.model_id == EMBEDDING_MODEL_ID and self.dim != DEFAULT_EMBEDDING_DIM:
+            object.__setattr__(self, "model_id", embedding_model_id(self.dim))
+
     def embed(self, text: str) -> List[float]:
         vector = [0.0] * self.dim
         features = _tokenize(text)
@@ -79,4 +87,4 @@ class LocalEmbeddingModel:
         return list(struct.unpack(f"<{count}f", payload[: count * 4]))
 
 
-__all__ = ["DEFAULT_EMBEDDING_DIM", "EMBEDDING_MODEL_ID", "LocalEmbeddingModel"]
+__all__ = ["DEFAULT_EMBEDDING_DIM", "EMBEDDING_MODEL_ID", "LocalEmbeddingModel", "embedding_model_id"]

package/lattice_brain/graph/projection.py

@@ -331,17 +331,31 @@ class KnowledgeGraphProjectionMixin:
         edge_id: Optional[str] = None,
         created_at: Optional[str] = None,
         strict: bool = False,
+        legacy_type: Optional[str] = None,
     ) -> None:
         if KGStoreV2 is None:
             if strict:
                 raise RuntimeError("Knowledge Graph v2 schema is unavailable")
             return
-        eid = (
-            edge_id or f"edge:{_sha256_text(f'{from_node}|{edge_type}|{to_node}')[:24]}"
-        )
+        explicit_legacy_type = legacy_type is not None
+        leg_type = legacy_type if explicit_legacy_type else edge_type
+        # Native canonical writes (and write-door dedupes) use legacy_type=''
+        # so (source,target,type) is the effective key.
+        # Import paths can pass distinct legacy_type to keep colliding legacy
+        # labels as separate rows (lossless for old data).
+        if leg_type and EdgeType is not None:
+            try:
+                if EdgeType.from_legacy(leg_type).value == leg_type:
+                    leg_type = ""
+            except ValueError:
+                pass
         norm_type = (
             EdgeType.from_legacy(edge_type).value if EdgeType is not None else edge_type
         )
+        if explicit_legacy_type and leg_type:
+            eid = f"edge:{_sha256_text(f'{from_node}|{norm_type}|{to_node}|{leg_type}')[:24]}"
+        else:
+            eid = edge_id or f"edge:{_sha256_text(f'{from_node}|{norm_type}|{to_node}')[:24]}"
         meta_str = metadata_json if metadata_json is not None else "{}"
         confidence = float(_safe_loads(meta_str).get("confidence", 1.0))
         try:
@@ -360,7 +374,7 @@ class KnowledgeGraphProjectionMixin:
                     from_node,
                     to_node,
                     norm_type,
-                    edge_type,
+                    leg_type,
                     float(weight),
                     confidence,
                     meta_str,
@@ -371,7 +385,7 @@ class KnowledgeGraphProjectionMixin:
             # (the UNIQUE upsert + weight=max alone would erase recurrence).
             row = conn.execute(
                 "SELECT id FROM edges_v2 WHERE source=? AND target=? AND type=? AND legacy_type=?",
-                (from_node, to_node, norm_type, edge_type),
+                (from_node, to_node, norm_type, leg_type),
             ).fetchone()
             if row is not None:
                 conn.execute(

package/lattice_brain/graph/provenance.py

@@ -292,10 +292,24 @@ class KnowledgeGraphProvenanceMixin:
             plan["dry_run"] = True
             return plan
 
-        if mode == "replace":
-            self.clear_all()
-
         with self._connect() as conn:
+            if mode == "replace":
+                # Keep replacement imports transactional. The old clear_all()
+                # path committed before the import started, so a malformed
+                # artifact could leave a cleared or partially rebuilt graph.
+                # These deletes roll back with the rest of the import.
+                for table in (
+                    "local_file_index",
+                    "knowledge_sources",
+                    "chunks",
+                    "edges",
+                    "nodes",
+                    "vector_embeddings",
+                ):
+                    conn.execute(f"DELETE FROM {table}")
+                if KGStoreV2 is not None:
+                    conn.execute("DELETE FROM edges_v2")
+                    conn.execute("DELETE FROM nodes_v2")
             for n in nodes:
                 self._upsert_node(
                     conn,
@@ -315,13 +329,21 @@ class KnowledgeGraphProvenanceMixin:
                     metadata=_safe_loads(c.get("metadata_json")),
                 )
             for e in edges:
+                e_meta = _safe_loads(e.get("metadata_json")) or {}
+                leg_label = e_meta.get("legacy_label")
+                if not leg_label:
+                    orig = e.get("type") or ""
+                    if orig:
+                        # preserve whatever label came from export (legacy or canon)
+                        leg_label = orig
                 self._upsert_edge(
                     conn,
                     e["from_node"],
                     e["to_node"],
                     e["type"],
                     weight=float(e.get("weight") or 1.0),
-                    metadata=_safe_loads(e.get("metadata_json")),
+                    metadata=e_meta,
+                    legacy_label=leg_label,
                 )
             for s in sources:
                 conn.execute(

package/lattice_brain/graph/write_master.py

@@ -73,20 +73,46 @@ class KnowledgeGraphWriteMixin:
         edge_type: str,
         weight: float = 1.0,
         metadata: Optional[Dict[str, Any]] = None,
+        *,
+        legacy_label: Optional[str] = None,
     ) -> str:
         # v4 write door: every new edge stores the canonical EdgeType value —
         # free-string types (e.g. '포함함', '언급함') are normalized here, so no
         # caller can mint new legacy taxonomy. The original label survives in
         # metadata.legacy_label for traceability.
+        #
+        # legacy_type in edges_v2:
+        #   - normal write door: synonyms always dedupe to ONE row with legacy_type=''
+        #   - import (passes legacy_label=) : preserves distinct legacy labels as
+        #     separate v2 rows (lossless collision for old artifacts / backfill)
+        passed_for_legacy = legacy_label or edge_type
         if EdgeType is not None:
             canonical = EdgeType.from_legacy(edge_type).value
-            if canonical != edge_type:
+            if canonical != edge_type or (legacy_label and legacy_label != canonical):
                 metadata = dict(metadata or {})
-                metadata.setdefault("legacy_label", edge_type)
+                ll = legacy_label or edge_type
+                if ll != canonical:
+                    metadata.setdefault("legacy_label", ll)
             edge_type = canonical
         edge_id = f"edge:{_sha256_text(f'{from_node}|{edge_type}|{to_node}')[:24]}"
         now = _now()
         meta_json = _json(metadata)  # canonical string shared with the projection
+        v2_legacy = ""
+        if passed_for_legacy and passed_for_legacy != edge_type:
+            if legacy_label is not None:
+                # explicit legacy_label from import path forces distinct legacy_type
+                # rows in v2 for collision preservation
+                v2_legacy = passed_for_legacy
+            else:
+                # normal write-door dedupes even for synonym labels (legacy_type='')
+                v2_legacy = ""
+        # v2 may use distinct eid when preserving legacy collisions (different from
+        # the canon-based edge_id used for legacy edges table which always dedupes)
+        v2_eid = (
+            f"edge:{_sha256_text(f'{from_node}|{edge_type}|{to_node}|{v2_legacy}')[:24]}"
+            if v2_legacy
+            else edge_id
+        )
         self._v2_project_edge(
             conn,
             from_node,
@@ -94,9 +120,10 @@ class KnowledgeGraphWriteMixin:
             edge_type,
             float(weight),
             meta_json,
-            edge_id=edge_id,
+            edge_id=v2_eid,
             created_at=now,
             strict=True,
+            legacy_type=v2_legacy,
         )
         conn.execute(
             """

package/lattice_brain/runtime/multi_agent.py

@@ -21,7 +21,7 @@ from typing import Any, Callable, Dict, List, Optional
 from .contracts import multi_agent_contract
 
 
-MULTI_AGENT_VERSION = "7.9.0"
+MULTI_AGENT_VERSION = "8.1.0"
 
 AGENT_ROLES = ("researcher", "planner", "executor", "reviewer", "release")
 CORE_PIPELINE = ("planner", "executor", "reviewer")

package/latticeai/__init__.py

@@ -1,3 +1,3 @@
 """Lattice AI - modular server package."""
 
-__version__ = "7.9.0"
+__version__ = "8.1.0"

package/latticeai/app_factory.py

@@ -77,8 +77,6 @@ def _build(config: "Optional[Config]" = None) -> Dict[str, Any]:
     import re
     import secrets
     import threading
-    import subprocess
-    import sys
     import time
     from pathlib import Path
 
@@ -195,8 +193,10 @@ def _build(config: "Optional[Config]" = None) -> Dict[str, Any]:
     )
     from latticeai.core.tool_registry import TOOL_CATALOG_BRIEF as _TOOL_CATALOG_BRIEF  # noqa: F401
     from latticeai.core.mcp_registry import (
-        _get_combined_registry,
-        _fetch_skills_marketplace, install_skill, SKILLS_DIR,
+        _fetch_skills_marketplace,
+        install_skill,
+        SKILLS_DIR,
+        create_mcp_install_state,
     )
     from latticeai.services.p_reinforce import PReinforceGardener
     from setup_wizard import get_recommendations, scan_environment
@@ -322,6 +322,15 @@ def _build(config: "Optional[Config]" = None) -> Dict[str, Any]:
     MCP_FILE = DATA_DIR / "mcp_installs.json"
     AUDIT_FILE = DATA_DIR / "audit_log.json"
     SSO_FILE = DATA_DIR / "sso_config.json"
+
+    # MCP state extracted to mcp_registry.create_mcp_install_state (server decomp)
+    _mcp_state = create_mcp_install_state(DATA_DIR)
+    load_mcp_installs = _mcp_state["load_mcp_installs"]
+    save_mcp_installs = _mcp_state["save_mcp_installs"]
+    mcp_public_item = _mcp_state["mcp_public_item"]
+    recommend_mcps = _mcp_state["recommend_mcps"]
+    install_mcp = _mcp_state["install_mcp"]
+
     # Resolve the configured embedding provider once at startup. Degrades to the
     # offline hash fallback when the requested provider is unavailable, while
     # recording the requested-vs-active provider for the Embeddings status surface.
@@ -546,136 +555,6 @@ def _build(config: "Optional[Config]" = None) -> Dict[str, Any]:
         with open(VPC_FILE, "w", encoding="utf-8") as f:
             json.dump(config, f, ensure_ascii=False, indent=2)
 
-    def load_mcp_installs() -> Dict:
-        if not os.path.exists(MCP_FILE):
-            return {"installed": {}, "updated_at": None}
-        try:
-            with open(MCP_FILE, "r", encoding="utf-8") as f:
-                data = json.load(f)
-            if "installed" not in data:
-                data["installed"] = {}
-            return data
-        except Exception as e:
-            logging.warning("load_mcp_installs failed: %s", e)
-            return {"installed": {}, "updated_at": None}
-
-    def save_mcp_installs(data: Dict):
-        data["updated_at"] = datetime.now().isoformat()
-        with open(MCP_FILE, "w", encoding="utf-8") as f:
-            json.dump(data, f, ensure_ascii=False, indent=2)
-
-    def mcp_public_item(item: Dict, installed_state: Dict) -> Dict:
-        state = installed_state.get(item["id"]) or {}
-        installed = item["install_mode"] in {"builtin", "bundled"} or bool(state.get("installed"))
-        connector_pending = item["install_mode"] == "connector" and not state.get("authenticated")
-        authenticated = item["install_mode"] != "connector" or bool(state.get("authenticated"))
-        return {
-            "id": item["id"],
-            "name": item["name"],
-            "category": item.get("category", ""),
-            "install_mode": item["install_mode"],
-            "description": item.get("description", ""),
-            "capabilities": item.get("capabilities", []),
-            "connector_url": item.get("connector_url"),
-            "external_url": item.get("external_url"),
-            "package": item.get("package"),
-            "homepage": item.get("homepage"),
-            "source": item.get("source", "local"),
-            "installed": installed,
-            "status": state.get("status") or ("active" if installed and not connector_pending else "needs_auth" if connector_pending else "available"),
-            "authenticated": authenticated,
-            "updated_at": state.get("updated_at"),
-        }
-
-    async def recommend_mcps(query: str, limit: int = 5) -> List[Dict]:
-        text = (query or "").lower()
-        installed = load_mcp_installs().get("installed", {})
-        registry = await _get_combined_registry()
-        scored = []
-        for item in registry:
-            score = 0
-            hits = []
-            for keyword in item.get("keywords", []):
-                if keyword.lower() in text:
-                    score += 3 if len(keyword) > 2 else 1
-                    hits.append(keyword)
-            # description 키워드 매칭 (remote 항목 보완)
-            if not hits and text:
-                desc_words = item.get("description", "").lower().split()
-                for word in text.split():
-                    if len(word) > 2 and word in desc_words:
-                        score += 1
-                        hits.append(word)
-            if item["id"] == "filesystem" and any(word in text for word in ["만들", "구현", "build", "deploy", "코드", "앱"]):
-                score += 2
-            if score:
-                public = mcp_public_item(item, installed)
-                public["score"] = score
-                public["matched_keywords"] = hits[:6]
-                scored.append(public)
-        if not scored:
-            fallback_ids = ["filesystem", "browser", "documents"]
-            scored = [
-                {**mcp_public_item(item, installed), "score": 1, "matched_keywords": []}
-                for item in registry
-                if item["id"] in fallback_ids
-            ]
-        return sorted(scored, key=lambda item: item["score"], reverse=True)[: max(1, min(limit, 24))]
-
-    async def install_mcp(mcp_id: str) -> Dict:
-        registry = await _get_combined_registry()
-        item = next((entry for entry in registry if entry["id"] == mcp_id), None)
-        if not item:
-            raise HTTPException(status_code=404, detail="MCP를 찾을 수 없습니다.")
-        data = load_mcp_installs()
-        state = data.setdefault("installed", {})
-        status = "active"
-        message = "MCP가 활성화되었습니다."
-        if item["install_mode"] == "connector":
-            status = "needs_auth"
-            message = "커넥터 인증이 필요합니다. Codex 앱의 connector 설정에서 계정을 연결하면 바로 사용할 수 있습니다."
-        elif item["install_mode"] == "pip":
-            packages = item.get("pip_packages") or []
-            for pkg in packages:
-                completed = subprocess.run(
-                    [sys.executable, "-m", "pip", "install", "--upgrade", pkg],
-                    capture_output=True, text=True, timeout=900, check=False,
-                )
-                if completed.returncode != 0:
-                    raise HTTPException(status_code=500, detail=completed.stderr[-2000:] or f"{pkg} 설치 실패")
-            message = f"필수 패키지 설치 완료: {', '.join(packages)}"
-        elif item["install_mode"] == "pypi":
-            pkg = item.get("package", "")
-            version = item.get("package_version")
-            pkg_str = f"{pkg}=={version}" if version else pkg
-            completed = subprocess.run(
-                [sys.executable, "-m", "pip", "install", pkg_str],
-                capture_output=True, text=True, timeout=300, check=False,
-            )
-            if completed.returncode != 0:
-                raise HTTPException(status_code=500, detail=completed.stderr[-2000:] or f"{pkg} 설치 실패")
-            message = f"pip 패키지 설치 완료: {pkg_str}"
-        elif item["install_mode"] == "npm":
-            pkg = item.get("package", "")
-            version = item.get("package_version")
-            pkg_str = f"{pkg}@{version}" if version else pkg
-            completed = subprocess.run(
-                ["npm", "install", "-g", pkg_str],
-                capture_output=True, text=True, timeout=300, check=False,
-            )
-            if completed.returncode != 0:
-                raise HTTPException(status_code=500, detail=completed.stderr[-2000:] or f"{pkg} 설치 실패")
-            message = f"npm 패키지 설치 완료: {pkg_str}"
-        state[mcp_id] = {
-            "installed": True,
-            "status": status,
-            "authenticated": item["install_mode"] != "connector",
-            "updated_at": datetime.now().isoformat(),
-        }
-        save_mcp_installs(data)
-        public = mcp_public_item(item, state)
-        public["message"] = message
-        return public
 
     _history_lock = threading.Lock()