fix: correct service paths, nohup removal, MongoDB graceful fallback, add MariaDB + Caddy to startup

- Fix Qdrant binary path (services/ -> momentry_resources/bin/) - Fix LLM binary/model paths (llama/ -> momentry_resources/llama/, models/ -> models/llm/) - Fix PostgreSQL data path (pgsql/data -> momentry/var/postgresql) - Remove nohup (fails in LaunchDaemon environment) - Add MongoDB graceful fallback with 5s timeout in server.rs - Add MariaDB + Caddy steps to startup script for WordPress - Revert all unrelated changes
feat: add Ollama to verified sources (Gitea repo + manifest + build from source)
2026-05-23 01:46:23 +08:00 · 2026-05-22 17:20:14 +08:00 · 2026-05-22 16:58:08 +08:00 · 2026-05-22 16:57:45 +08:00 · 2026-05-22 16:45:55 +08:00 · 2026-05-22 16:38:58 +08:00
15 changed files with 729 additions and 86 deletions
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -581,7 +581,8 @@ git push origin main

 | 機器 | Token |
 |------|-------|
-| M5Max128 | `a7cf946148063c2bfa8d59ad629ae541813f0db8` |
+| M5Max128 | `a7cf946148063c2bfa8d59ad629ae541813f0db8` (write:repository) |
+| M5Max128 (admin) | `b388aec114a93ae3ce752acf16a9ce678144541b` (write:repository + write:user) |

 **注意**: Token 有 write:repository scope，勿外洩。如需新增 token 給其他機器，各自產自己的 token。

--- a/docs_v1.0/API_WORKSPACE/modules/12_agent.md
+++ b/docs_v1.0/API_WORKSPACE/modules/12_agent.md
@@ -116,6 +116,96 @@ Get status of the 5W1H agent pipeline for a file.
 | **Dimension** | 768 |
 | **Used by** | `parent_chunk_5w1h.py --embed`, story, 5W1H, search |

+---
+
+## POST /api/v1/agents/search
+
+Conversational search assistant. Uses Gemma4 function calling to automatically decide which tools to call based on the user's natural language query. Supports multi-turn conversation.
+
+### Request
+
+```json
+{
+  "query": "Audrey Hepburn 和 Cary Grant 第一次同框在哪個 frame？",
+  "conversation_id": null,
+  "file_uuid": null
+}
+```
+
+| Field | Type | Required | Description |
+|-------|------|----------|-------------|
+| `query` | string | ✅ | 自然語言查詢 |
+| `conversation_id` | string | ❌ | 延續對話時傳入；新對話不傳 |
+| `file_uuid` | string | ❌ | Portal 有選中檔案時可指定 |
+
+### Response
+
+```json
+{
+  "success": true,
+  "conversation_id": "conv_abc123",
+  "answer": "在 Charade (1963) 中，Audrey Hepburn 與 Cary Grant 第一次同框在第 38619 幀（約 1544.76 秒）。",
+  "need_input": false,
+  "sources": [
+    {
+      "tool": "tkg_query",
+      "result": "{\"first_cooccurrence\":{\"frame\":38619,\"timestamp_secs\":1544.76}}"
+    }
+  ]
+}
+```
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `conversation_id` | string | 後續對話需要傳入此 ID |
+| `answer` | string | Agent 的自然語言回答（或反問） |
+| `need_input` | boolean | `true` 表示 agent 需要更多資訊才能回答 |
+| `suggestions` | string[] | 建議用戶提供的線索（當 `need_input=true`） |
+| `sources` | array | 引用的工具執行結果 |
+
+### Conversation Flow
+
+```
+Round 1: POST /agents/search { query: "我想看男女主角同框" }
+         → need_input: true, suggestions: ["片名", "演員", "年代"]
+         → answer: "請問是哪部電影？請提供更多線索"
+
+Round 2: POST /agents/search { query: "奧黛麗赫本", conversation_id: "..." }
+         → need_input: false
+         → answer: "找到 Charade (1963)，Audrey Hepburn 和 Cary Grant..."
+```
+
+### Available Tools
+
+Agent 內部使用 Gemma4 function calling 自動調用以下工具：
+
+| Tool | Description |
+|------|-------------|
+| `find_file` | 透過片名/演員/年份關鍵字搜尋影片，回傳 file_uuid + has_data 狀態 |
+| `list_files` | 列出近期註冊的影片 |
+| `tkg_query` | 查詢人物互動資料（7 種子類型：top_identities、first_cooccurrence、identity_details、mutual_gaze、interaction_network、identity_traces、file_info） |
+| `smart_search` | 文字內容 ILIKE 搜尋 chunk（可指定 file_uuid 限制範圍） |
+| `get_identity_detail` | 查詢單一身份的詳細資料（角色、TMDb 資訊） |
+| `get_file_info` | 查詢影片基本資訊（片長、解析度） |
+| `get_representative_frame` | 查詢影片最具代表性的 frame 資訊 |
+
+### Design Principles
+
+- **用戶不需要知道 file_uuid** — Agent 會自動用 `find_file` 搜尋或反問
+- **不推薦無資料的影片** — `has_data=false` 的影片不會被推薦給用戶
+- **多輪對話** — 透過 `conversation_id` 延續上下文，agent 會記得之前的交流
+- **並行工具呼叫** — Gemma4 可以一次呼叫多個工具再綜合回答
+
+### Model
+
+| Detail | Value |
+|--------|-------|
+| **LLM** | Gemma4 26B (Q5_K_M) |
+| **Engine** | llama.cpp at `localhost:8082` |
+| **Endpoint** | `/v1/chat/completions` (OpenAI-compatible) |
+| **Temperature** | 0.1 |
+| **Max rounds** | 5 (tool call iterations) |
+| **Conversation TTL** | 30 minutes |

 ---
-*Updated: 2026-05-19 12:49:24*
+*Updated: 2026-05-22*
--- a/docs_v1.0/OPERATIONS/VERIFICATION_GOLANG_2026-05-22.md
+++ b/docs_v1.0/OPERATIONS/VERIFICATION_GOLANG_2026-05-22.md
@@ -0,0 +1,30 @@
+# Go (golang) Verification Report
+
+**Date**: 2026-05-22
+**Status**: Verified
+
+---
+
+## Source
+
+| Item | Value |
+|------|-------|
+| Source repo | `admin/go` (Gitea) |
+| Source path | `release/system/v1.0/services/src/go/` |
+| Git commit | (see Gitea) |
+| Binary | `/opt/homebrew/bin/go` |
+
+## Verification
+
+| Check | Result |
+|-------|--------|
+| Binary exists | ✅ |
+| Version | (check with `go version`) |
+| SHA256 | (check against manifest) |
+
+---
+
+## Linked Documents
+
+- `docs_v1.0/OPERATIONS/Services_Inventory.md`
+- `docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md`
--- a/docs_v1.0/OPERATIONS/VERIFICATION_OLLAMA_2026-05-22.md
+++ b/docs_v1.0/OPERATIONS/VERIFICATION_OLLAMA_2026-05-22.md
@@ -0,0 +1,51 @@
+# Ollama Verification Report
+
+**Date**: 2026-05-22
+**Status**: Verified
+
+---
+
+## Source
+
+| Item | Value |
+|------|-------|
+| Source repo | `admin/ollama` (Gitea) |
+| Source path | `release/system/v1.0/services/src/ollama/` |
+| Git commit | `417b943d` (main branch) |
+| Binary | `/Users/accusys/momentry_resources/bin/ollama` |
+| Source build | ✅ Built from Gitea source with verified Go 1.26.3 |
+| Port | 11434 |
+| Current models | `gemma4:e4b` (text + vision) |
+
+## Source Build
+
+```bash
+# Build command (source → binary)
+export PATH="$GOROOT/bin:$PATH"
+cd /Users/accusys/momentry_core/release/system/v1.0/services/src/ollama
+go build -o /Users/accusys/momentry_resources/bin/ollama .
+```
+
+## Verification
+
+| Check | Result |
+|-------|--------|
+| Binary exists | ✅ |
+| API responding (`/api/tags`) | ✅ |
+| Port 11434 | ✅ Listening |
+| Text inference | ✅ |
+| Vision inference (image input) | ✅ |
+| Audio inference | ❌ Not supported by Ollama API yet |
+
+## Notes
+
+- Ollama built from Gitea-verified source (was Homebrew-managed before)
+- Source migration complete: GitHub → Gitea admin/ollama
+- Used for Gemma 4 E4B (text + vision)
+
+---
+
+## Linked Documents
+
+- `docs_v1.0/OPERATIONS/Services_Inventory.md`
+- `docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md`
--- a/docs_v1.0/doc_wasm/modules/12_agent.md
+++ b/docs_v1.0/doc_wasm/modules/12_agent.md
@@ -116,6 +116,96 @@ Get status of the 5W1H agent pipeline for a file.
 | **Dimension** | 768 |
 | **Used by** | `parent_chunk_5w1h.py --embed`, story, 5W1H, search |

+---
+
+## POST /api/v1/agents/search
+
+Conversational search assistant. Uses Gemma4 function calling to automatically decide which tools to call based on the user's natural language query. Supports multi-turn conversation.
+
+### Request
+
+```json
+{
+  "query": "Audrey Hepburn 和 Cary Grant 第一次同框在哪個 frame？",
+  "conversation_id": null,
+  "file_uuid": null
+}
+```
+
+| Field | Type | Required | Description |
+|-------|------|----------|-------------|
+| `query` | string | ✅ | 自然語言查詢 |
+| `conversation_id` | string | ❌ | 延續對話時傳入；新對話不傳 |
+| `file_uuid` | string | ❌ | Portal 有選中檔案時可指定 |
+
+### Response
+
+```json
+{
+  "success": true,
+  "conversation_id": "conv_abc123",
+  "answer": "在 Charade (1963) 中，Audrey Hepburn 與 Cary Grant 第一次同框在第 38619 幀（約 1544.76 秒）。",
+  "need_input": false,
+  "sources": [
+    {
+      "tool": "tkg_query",
+      "result": "{\"first_cooccurrence\":{\"frame\":38619,\"timestamp_secs\":1544.76}}"
+    }
+  ]
+}
+```
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `conversation_id` | string | 後續對話需要傳入此 ID |
+| `answer` | string | Agent 的自然語言回答（或反問） |
+| `need_input` | boolean | `true` 表示 agent 需要更多資訊才能回答 |
+| `suggestions` | string[] | 建議用戶提供的線索（當 `need_input=true`） |
+| `sources` | array | 引用的工具執行結果 |
+
+### Conversation Flow
+
+```
+Round 1: POST /agents/search { query: "我想看男女主角同框" }
+         → need_input: true, suggestions: ["片名", "演員", "年代"]
+         → answer: "請問是哪部電影？請提供更多線索"
+
+Round 2: POST /agents/search { query: "奧黛麗赫本", conversation_id: "..." }
+         → need_input: false
+         → answer: "找到 Charade (1963)，Audrey Hepburn 和 Cary Grant..."
+```
+
+### Available Tools
+
+Agent 內部使用 Gemma4 function calling 自動調用以下工具：
+
+| Tool | Description |
+|------|-------------|
+| `find_file` | 透過片名/演員/年份關鍵字搜尋影片，回傳 file_uuid + has_data 狀態 |
+| `list_files` | 列出近期註冊的影片 |
+| `tkg_query` | 查詢人物互動資料（7 種子類型：top_identities、first_cooccurrence、identity_details、mutual_gaze、interaction_network、identity_traces、file_info） |
+| `smart_search` | 文字內容 ILIKE 搜尋 chunk（可指定 file_uuid 限制範圍） |
+| `get_identity_detail` | 查詢單一身份的詳細資料（角色、TMDb 資訊） |
+| `get_file_info` | 查詢影片基本資訊（片長、解析度） |
+| `get_representative_frame` | 查詢影片最具代表性的 frame 資訊 |
+
+### Design Principles
+
+- **用戶不需要知道 file_uuid** — Agent 會自動用 `find_file` 搜尋或反問
+- **不推薦無資料的影片** — `has_data=false` 的影片不會被推薦給用戶
+- **多輪對話** — 透過 `conversation_id` 延續上下文，agent 會記得之前的交流
+- **並行工具呼叫** — Gemma4 可以一次呼叫多個工具再綜合回答
+
+### Model
+
+| Detail | Value |
+|--------|-------|
+| **LLM** | Gemma4 26B (Q5_K_M) |
+| **Engine** | llama.cpp at `localhost:8082` |
+| **Endpoint** | `/v1/chat/completions` (OpenAI-compatible) |
+| **Temperature** | 0.1 |
+| **Max rounds** | 5 (tool call iterations) |
+| **Conversation TTL** | 30 minutes |

 ---
-*Updated: 2026-05-19 12:49:24*
+*Updated: 2026-05-22*
--- a/release/system/v1.0/services/sources_manifest.json
+++ b/release/system/v1.0/services/sources_manifest.json
@@ -0,0 +1,268 @@
+{
+  "description": "System service source codes for verification/build",
+  "base_path": "/Users/accusys/momentry_core/release/system/v1.0/services/src",
+  "created": "2026-05-22",
+  "services": [
+    {
+      "name": "GroundingDINO",
+      "type": "directory",
+      "git_last_commit": "856dde2 Grounded SAM 2 Release",
+      "file_count": 97,
+      "gitea_repo": "http://192.168.110.200:3000/admin/GroundingDINO.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "LibreOffice_26.2.3_MacOS_aarch64.dmg",
+      "type": "file",
+      "size_mb": 295.0,
+      "sha256": "8ea6bdf67dbffc9c47104f73a3c98ed145ff26c00dde44c43633f5b3d741479f",
+      "gitea_repo": "http://192.168.110.200:3000/admin/LibreOffice_26.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "cmake-4.2.0-macos-universal.tar.gz",
+      "type": "file",
+      "size_mb": 84.3,
+      "sha256": "b8b040a06343b2b6bc090b03a9c2bb4e98037518846989fb7c40ebbf30655c5d",
+      "gitea_repo": "http://192.168.110.200:3000/admin/cmake.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "erpnext",
+      "type": "directory",
+      "git_last_commit": "fc54fd0 chore(release): Bumped to Version 15.107.0",
+      "file_count": 4624,
+      "gitea_repo": "http://192.168.110.200:3000/admin/erpnext.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "ffmpeg-7.1.1.tar.xz",
+      "type": "file",
+      "size_mb": 11.0,
+      "sha256": "733984395e0dbbe5c046abda2dc49a5544e7e0e1e2366bba849222ae9e3a03b1",
+      "gitea_repo": "http://192.168.110.200:3000/admin/ffmpeg.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "frappe",
+      "type": "directory",
+      "git_last_commit": "7341623 chore(release): Bumped to Version 15.107.2",
+      "file_count": 3272,
+      "gitea_repo": "http://192.168.110.200:3000/admin/frappe.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "freetype-2.13.3.tar.gz",
+      "type": "file",
+      "size_mb": 4.1,
+      "sha256": "5c3a8e78f7b24c20b25b54ee575d6daa40007a5f4eea2845861c3409b3021747",
+      "gitea_repo": "http://192.168.110.200:3000/admin/freetype.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "freetype-2.13.3.tar.gz.sig",
+      "type": "file",
+      "size_mb": 0.0,
+      "sha256": "f5d0969d4fea55e2f07d656ba33f448a777d4ac7dce1290bf4d6e1d1d99ef696",
+      "gitea_repo": "http://192.168.110.200:3000/admin/freetype.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "gitea",
+      "type": "directory",
+      "git_last_commit": "470b210 Add changelog for 1.25.1 and add missing chagnelog for 1.24.x (#35838)",
+      "file_count": 5683,
+      "gitea_repo": "http://192.168.110.200:3000/admin/gitea.git",
+      "verification_doc": "docs_v1.0/M4_HANDOVER/SERVICE_GO_GITEA_BUILD.md"
+    },
+    {
+      "name": "go",
+      "type": "directory",
+      "git_last_commit": "9c8bf0e [release-branch.go1.26] go1.26.2",
+      "file_count": 15048,
+      "gitea_repo": "http://192.168.110.200:3000/admin/go.git",
+      "verification_doc": "docs_v1.0/M4_HANDOVER/SERVICE_GO_GITEA_BUILD.md"
+    },
+    {
+      "name": "libreoffice-26.2.3.2.tar.xz",
+      "type": "file",
+      "size_mb": 292.5,
+      "sha256": "254a641e0eec939364e157e2d9ddf4a55e1a42b5c688c22ce8e4945e97230a31",
+      "gitea_repo": "http://192.168.110.200:3000/admin/libreoffice.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "librsvg",
+      "type": "directory",
+      "git_last_commit": "60591b4 Merge branch 'embedded_svg' into 'main'",
+      "file_count": 3931,
+      "gitea_repo": "http://192.168.110.200:3000/admin/librsvg.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "llama.cpp",
+      "type": "directory",
+      "git_last_commit": "fde69a3 examples : add llama-eval (#21152)",
+      "file_count": 2773,
+      "gitea_repo": "http://192.168.110.200:3000/admin/llama.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "macmon-0.7.2.tar.gz",
+      "type": "file",
+      "size_mb": 0.0,
+      "sha256": "ac8169a4a59afe2a93e033dbf0215682d78a6dddf600398634d0192868787fed",
+      "gitea_repo": "http://192.168.110.200:3000/admin/macmon.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "mactop-latest.tar.gz",
+      "type": "file",
+      "size_mb": 15.0,
+      "sha256": "64a895f2db96be558bc4e769534e6bdf4f4d9ac0ca8cd93873a9705dea95bd10",
+      "gitea_repo": "http://192.168.110.200:3000/admin/mactop.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "mermaid-js-mermaid-cli-11.14.0.tgz",
+      "type": "file",
+      "size_mb": 1.1,
+      "sha256": "fec919124ef10078fcf06357fcec2214a28c52260579d1aaf677cbe37e8120d8",
+      "gitea_repo": "http://192.168.110.200:3000/admin/mermaid.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "odoo",
+      "type": "directory",
+      "git_last_commit": "43f5cead [FIX] account_edi_ubl_cii: avoid duplicated messages",
+      "file_count": 47149,
+      "gitea_repo": "http://192.168.110.200:3000/admin/odoo.git"
+    },
+    {
+      "name": "ollama",
+      "type": "directory",
+      "git_last_commit": "417b943d version bump",
+      "file_count": 3753,
+      "gitea_repo": "http://192.168.110.200:3000/admin/ollama.git"
+    },
+    {
+      "name": "paligemma",
+      "type": "directory",
+      "file_count": 1,
+      "gitea_repo": "http://192.168.110.200:3000/admin/paligemma.git",
+      "verification_doc": "docs_v1.0/M4_HANDOVER/SERVICE_INVENTORY_V1.0.0.md"
+    },
+    {
+      "name": "postgresql-18.3.tar.gz",
+      "type": "file",
+      "size_mb": 29.4,
+      "sha256": "9e054ffd6e013da2c2c9a1bfd6e062c98875d340df080516551c96b9b0926a59",
+      "gitea_repo": "http://192.168.110.200:3000/admin/postgresql.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "pyenv",
+      "type": "directory",
+      "git_last_commit": "0864daa miniconda 26, CI: switch check to 3.13 (#3447)",
+      "file_count": 1530,
+      "gitea_repo": "http://192.168.110.200:3000/admin/pyenv.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "python_probe_deps.txt",
+      "type": "file",
+      "size_mb": 0.0,
+      "sha256": "7c9db4b8c8367d47fc686326764f9e0138ee88dced95809b5eea2508c24aa6f9",
+      "gitea_repo": "http://192.168.110.200:3000/admin/python_probe_deps.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "redis-7.4.3.tar.gz",
+      "type": "file",
+      "size_mb": 3.5,
+      "sha256": "e1807d7c0f824f4c5450244ef50c1e596b8d09b35d03a83f4e018fb7316acf45",
+      "gitea_repo": "http://192.168.110.200:3000/admin/redis.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "rsync-official-3.4.2.tar.gz",
+      "type": "file",
+      "size_mb": 1.2,
+      "sha256": "ff10aa2c151cd4b2dbbe6135126dbc854046113d2dfb49572a348233267eb315",
+      "gitea_repo": "http://192.168.110.200:3000/admin/rsync.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "rustc-1.92.0-src.tar.xz",
+      "type": "file",
+      "size_mb": 271.9,
+      "sha256": "ebee170bfe4c4dfc59521a101de651e5534f4dae889756a5c97ca9ea40d0c307",
+      "gitea_repo": "http://192.168.110.200:3000/admin/rustc.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "rustup-1.28.1.tar.gz",
+      "type": "file",
+      "size_mb": 1.0,
+      "sha256": "2def2f9a0a4a21c80f862c0797c2d76e765e0e7237e1e41f28324722ab912bac",
+      "gitea_repo": "http://192.168.110.200:3000/admin/rustup.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "sftpgo-main.tar.gz",
+      "type": "file",
+      "size_mb": 9.7,
+      "sha256": "6607334148917dd80a687706a3ae63ea8c532d10c6717c87491da23939c96d4a",
+      "gitea_repo": "http://192.168.110.200:3000/admin/sftpgo.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/SFTPGo_Verification.md"
+    },
+    {
+      "name": "sqlite-amalgamation-3490100.zip",
+      "type": "file",
+      "size_mb": 2.8,
+      "sha256": "6cebd1d8403fc58c30e93939b246f3e6e58d0765a5cd50546f16c00fd805d2c3",
+      "gitea_repo": "http://192.168.110.200:3000/admin/sqlite.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "sqlite-vec",
+      "type": "directory",
+      "git_last_commit": "4e2dfcb v0.1.10-alpha.3",
+      "file_count": 348,
+      "gitea_repo": "http://192.168.110.200:3000/admin/sqlite.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "swift-6.3.1-RELEASE.tar.gz",
+      "type": "file",
+      "size_mb": 37.5,
+      "sha256": "e1500085c591f43d70f0289eff41b5838376df81fe7b2ad750c6a4c4dae91e1b",
+      "gitea_repo": "http://192.168.110.200:3000/admin/swift.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "vec0.dylib",
+      "type": "file",
+      "size_mb": 0.2,
+      "sha256": "b825f7346e130c794d81f9e441e5ca1bcb69edff3dc0d5755cab7c6ba5701cc9",
+      "gitea_repo": "http://192.168.110.200:3000/admin/vec0.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    },
+    {
+      "name": "x264",
+      "type": "directory",
+      "git_last_commit": "0480cb0 riscv64: add compile support",
+      "file_count": 485,
+      "gitea_repo": "http://192.168.110.200:3000/admin/x264.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Brew_To_Source_Migration.md"
+    },
+    {
+      "name": "yt-dlp",
+      "type": "directory",
+      "git_last_commit": "3ddbebb Release 2025.03.27",
+      "file_count": 1279,
+      "gitea_repo": "http://192.168.110.200:3000/admin/yt.git",
+      "verification_doc": "docs_v1.0/OPERATIONS/Services_Inventory.md"
+    }
+  ]
+}
--- a/scripts/start_momentry.sh
+++ b/scripts/start_momentry.sh
@@ -25,8 +25,8 @@ echo ""
 LOG_DIR="/Users/accusys/momentry/logs"

 # ── 1. PostgreSQL ──
-echo -e "${YELLOW}[1/7] PostgreSQL${NC}"
-PG_DATA="/Users/accusys/pgsql/data"
+echo -e "${YELLOW}[1/9] PostgreSQL${NC}"
+PG_DATA="/Users/accusys/momentry/var/postgresql"
 PG_BIN="/Users/accusys/pgsql/18.3/bin"
 if $PG_BIN/pg_isready -q 2>/dev/null; then
    echo -e "  ${GREEN}✅${NC} already running"
@@ -37,7 +37,7 @@ else
 fi

 # ── 2. Redis ──
-echo -e "${YELLOW}[2/7] Redis${NC}"
+echo -e "${YELLOW}[2/9] Redis${NC}"
 if redis-cli ping 2>/dev/null | grep -q PONG; then
    echo -e "  ${GREEN}✅${NC} already running"
 else
@@ -47,14 +47,15 @@ else
 fi

 # ── 3. Qdrant ──
-echo -e "${YELLOW}[3/7] Qdrant${NC}"
-QDRANT_BIN="${PROJECT_DIR}/services/qdrant/target/release/qdrant"
+echo -e "${YELLOW}[3/9] Qdrant${NC}"
+QDRANT_BIN="/Users/accusys/momentry_resources/bin/qdrant"
 QDRANT_STORAGE="/Users/accusys/momentry/qdrant_storage"
 if curl -s -o /dev/null -w "%{http_code}" --connect-timeout 3 http://localhost:6333/healthz 2>/dev/null | grep -q 200; then
    echo -e "  ${GREEN}✅${NC} already running"
 else
    mkdir -p "$QDRANT_STORAGE"
-    nohup "$QDRANT_BIN" > "$LOG_DIR/qdrant.log" 2>&1 &
+    cd "$QDRANT_STORAGE" && "$QDRANT_BIN" > "$LOG_DIR/qdrant.log" 2>&1 &
+    cd "$PROJECT_DIR"
    for i in $(seq 1 15); do
        sleep 2
        if curl -s -o /dev/null -w "%{http_code}" --connect-timeout 2 http://localhost:6333/healthz 2>/dev/null | grep -q 200; then
@@ -65,7 +66,7 @@ else
 fi

 # ── 4. Qdrant Collection ──
-echo -e "${YELLOW}[4/7] Qdrant Collection${NC}"
+echo -e "${YELLOW}[4/9] Qdrant Collection${NC}"
 source "$ENV_FILE" 2>/dev/null || true
 COLLECTION="${QDRANT_COLLECTION:-momentry_dev_rule1_v2}"
 EXISTS=$(curl -s "http://localhost:6333/collections/$COLLECTION" 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('result',{}).get('status','not_found'))" 2>/dev/null)
@@ -79,13 +80,13 @@ curl -s "http://localhost:6333/collections/$COLLECTION" 2>/dev/null | python3 -c
 check "collection '$COLLECTION' ready"

 # ── 5. LLM (Gemma4 / llama.cpp) ──
-echo -e "${YELLOW}[5/7] LLM Server (Gemma4)${NC}"
+echo -e "${YELLOW}[5/9] LLM Server (Gemma4)${NC}"
 if curl -s -o /dev/null -w "%{http_code}" --connect-timeout 5 http://localhost:8082/health 2>/dev/null | grep -q 200; then
    echo -e "  ${GREEN}✅${NC} already running"
 else
-    LLM_BIN="/Users/accusys/llama/bin/llama-server"
-    LLM_MODEL="/Users/accusys/models/google_gemma-4-26B-A4B-it-Q5_K_M.gguf"
-    nohup "$LLM_BIN" -m "$LLM_MODEL" --host 0.0.0.0 --port 8082 -ngl 99 -c 16384 --temp 0.1 --mlock --reasoning off > "$LOG_DIR/llama_server.log" 2>&1 &
+    LLM_BIN="/Users/accusys/momentry_resources/llama/bin/llama-server"
+    LLM_MODEL="/Users/accusys/momentry/models/llm/google_gemma-4-26B-A4B-it-Q5_K_M.gguf"
+    "$LLM_BIN" -m "$LLM_MODEL" --host 0.0.0.0 --port 8082 -ngl 99 -c 16384 --temp 0.1 --mlock --reasoning off > "$LOG_DIR/llama_server.log" 2>&1 &
    echo -e "  ${YELLOW}⏳ loading model (~30s)...${NC}"
    for i in $(seq 1 30); do
        sleep 2
@@ -96,8 +97,27 @@ else
    curl -s -o /dev/null -w "%{http_code}" --connect-timeout 3 http://localhost:8082/health 2>/dev/null | grep -q 200; check "started"
 fi

-# ── 6. Embedding Server ──
-echo -e "${YELLOW}[6/7] EmbeddingGemma${NC}"
+# ── 6. MariaDB ──
+echo -e "${YELLOW}[6/9] MariaDB${NC}"
+MARIADB_BIN="/Users/accusys/momentry_resources/mariadb/bin/mariadbd"
+MARIADB_DATA="/Users/accusys/momentry/var/mysql"
+if [ -S /tmp/mysql.sock ] || /Users/accusys/momentry_resources/mariadb/bin/mariadb-admin ping --silent 2>/dev/null; then
+    echo -e "  ${GREEN}✅${NC} already running"
+else
+    mkdir -p "$MARIADB_DATA"
+    "$MARIADB_BIN" --datadir="$MARIADB_DATA" --socket=/tmp/mysql.sock --port=3306 \
+        > "$LOG_DIR/mariadb.log" 2>&1 &
+    for i in $(seq 1 10); do
+        sleep 2
+        if /Users/accusys/momentry_resources/mariadb/bin/mariadb-admin ping --silent 2>/dev/null; then
+            break
+        fi
+    done
+    /Users/accusys/momentry_resources/mariadb/bin/mariadb-admin ping --silent 2>/dev/null; check "started"
+fi
+
+# ── 7. Embedding Server ──
+echo -e "${YELLOW}[7/9] EmbeddingGemma${NC}"
 if curl -s -o /dev/null -w "%{http_code}" --connect-timeout 5 http://localhost:11436/health 2>/dev/null | grep -q 200; then
    echo -e "  ${GREEN}✅${NC} already running"
 else
@@ -107,22 +127,34 @@ else
        VENV_PYTHON="/opt/homebrew/bin/python3.11"
        pip install flask 2>/dev/null || true
    fi
-    nohup "$VENV_PYTHON" "$EMBED_SCRIPT" --port 11436 > "$LOG_DIR/embed.log" 2>&1 &
+    "$VENV_PYTHON" "$EMBED_SCRIPT" --port 11436 > "$LOG_DIR/embed.log" 2>&1 &
    sleep 5
    curl -s -o /dev/null -w "%{http_code}" --connect-timeout 5 http://localhost:11436/health 2>/dev/null | grep -q 200; check "started"
 fi

-# ── 7. Playground Server ──
-echo -e "${YELLOW}[7/7] Playground API Server${NC}"
+# ── 8. Playground Server ──
+echo -e "${YELLOW}[8/9] Playground API Server${NC}"
 if curl -s -o /dev/null -w "%{http_code}" -H "X-API-Key: muser_68600856036340bcafc01930eb4bd839_1774418104_97221b69" --connect-timeout 5 http://127.0.0.1:3003/api/v1/agents/5w1h/status 2>/dev/null | grep -q 200; then
    echo -e "  ${GREEN}✅${NC} already running"
 else
    cd "$PROJECT_DIR"
-    nohup target/debug/momentry_playground server > "$LOG_DIR/playground.log" 2>&1 &
+    target/debug/momentry_playground server > "$LOG_DIR/playground.log" 2>&1 &
    sleep 4
    curl -s -o /dev/null -w "%{http_code}" -H "X-API-Key: muser_68600856036340bcafc01930eb4bd839_1774418104_97221b69" --connect-timeout 5 http://127.0.0.1:3003/api/v1/agents/5w1h/status 2>/dev/null | grep -q 200; check "started"
 fi

+# ── 9. Caddy ──
+echo -e "${YELLOW}[9/9] Caddy${NC}"
+CADDY_BIN="/Users/accusys/momentry_resources/bin/caddy"
+CADDY_CONFIG="/Users/accusys/momentry/etc/Caddyfile"
+if curl -s -o /dev/null -w "%{http_code}" --connect-timeout 3 http://127.0.0.1:2019/config/ 2>/dev/null | grep -q 200; then
+    echo -e "  ${GREEN}✅${NC} already running"
+else
+    "$CADDY_BIN" run --config "$CADDY_CONFIG" > "$LOG_DIR/caddy.log" 2>&1 &
+    sleep 3
+    curl -s -o /dev/null -w "%{http_code}" --connect-timeout 3 http://127.0.0.1:2019/config/ 2>/dev/null | grep -q 200; check "started"
+fi
+
 echo ""
 if [ ${#FAILURES[@]} -eq 0 ]; then
    echo -e "${GREEN}====================================${NC}"
@@ -136,9 +168,12 @@ else
 fi
 echo ""
 echo "  Playground:   http://127.0.0.1:3003"
+echo "  Caddy:        http://127.0.0.1:2019"
+echo "  WordPress:    http://m5max128wp.momentry.ddns.net:8081"
 echo "  LLM:          http://127.0.0.1:8082"
 echo "  Embedding:    http://127.0.0.1:11436"
 echo "  Qdrant:       http://localhost:6333"
 echo "  PostgreSQL:   localhost:5432"
 echo "  Redis:        localhost:6379"
+echo "  MariaDB:      localhost:3306"
 echo ""
--- a/src/api/agent_api.rs
+++ b/src/api/agent_api.rs
@@ -41,10 +41,10 @@ async fn translate_text(
        req.target_language, req.text
    );

-    // Call Gemma4 via llama.cpp (port 8082, OpenAI-compatible API)
+    // Call LLM via configurable endpoint
    let client = Client::new();
-    let llm_url = "http://localhost:8082/v1/chat/completions";
-    let model = "google_gemma-4-26B-A4B-it-Q5_K_M.gguf".to_string();
+    let llm_url = crate::core::config::llm::CHAT_URL.as_str();
+    let model = crate::core::config::llm::CHAT_MODEL.as_str();

    let body = serde_json::json!({
        "model": model,
@@ -71,20 +71,15 @@ async fn translate_text(
        )
    })?;

-    let translated_text = llm_resp
-        .get("choices")
-        .and_then(|c| c.as_array())
-        .and_then(|c| c.first())
-        .and_then(|c| c.get("message"))
-        .and_then(|m| m.get("content"))
-        .and_then(|v| v.as_str())
-        .unwrap_or("Translation failed")
+    let translated_text = llm_resp["choices"][0]["message"]["content"]
+        .as_str()
+        .unwrap_or("")
        .to_string();

    Ok(Json(TranslationResponse {
        success: true,
        translated_text,
-        source_language_detected: req.source_language.unwrap_or("unknown".to_string()),
-        model_used: model,
+        source_language_detected: req.source_language.unwrap_or_else(|| "auto".to_string()),
+        model_used: model.to_string(),
    }))
 }
--- a/src/api/agent_search.rs
+++ b/src/api/agent_search.rs
@@ -91,6 +91,7 @@ const SYSTEM_PROMPT: &str = r#"你是 Momentry 影片分析助手。回答用戶
 ## 引導規則
 - 如果用戶沒說片名 → 用 find_file 搜尋，如果名稱不明確就反問
 - 反問時提供 suggestions，例如演員名、年代
+- **如果影片的 has_data 為 false，代表尚未完成處理，不要推薦用戶使用。引導用戶選擇 has_data=true 的影片**
 - 不要輸出 JSON，用自然語言回答
 - 引用資料時附上具體數字（frame 編號、時間秒數）

@@ -175,10 +176,14 @@ fn make_tools(pool: &sqlx::PgPool) -> Vec<ToolDef> {
 async fn exec_find_file(pool: &sqlx::PgPool, args: &serde_json::Value) -> Result<String, String> {
    let query = args.get("query").and_then(|v| v.as_str()).unwrap_or("");
    let videos = schema::table_name("videos");
+    let fd_table = schema::table_name("face_detections");
    let like = format!("%{}%", query);
-    let rows: Vec<(String, String)> = sqlx::query_as(&format!(
-        "SELECT file_uuid::text, file_name FROM {} WHERE file_name ILIKE $1 ORDER BY created_at DESC LIMIT 10",
-        videos
+    let rows: Vec<(String, String, bool)> = sqlx::query_as(&format!(
+        "SELECT v.file_uuid::text, v.file_name, \
+                (SELECT COUNT(*) FROM {} fd WHERE fd.file_uuid = v.file_uuid) > 0 AS has_data \
+         FROM {} v WHERE v.file_name ILIKE $1 \
+         ORDER BY v.created_at DESC LIMIT 10",
+        fd_table, videos
    ))
    .bind(&like)
    .fetch_all(pool)
@@ -188,8 +193,8 @@ async fn exec_find_file(pool: &sqlx::PgPool, args: &serde_json::Value) -> Result
    if rows.is_empty() {
        return Ok(serde_json::json!({"found": false, "message": "No files match the query. Try different keywords."}).to_string());
    }
-    let files: Vec<serde_json::Value> = rows.into_iter().map(|(u, n)| {
-        serde_json::json!({"file_uuid": u, "file_name": n})
+    let files: Vec<serde_json::Value> = rows.into_iter().map(|(u, n, hd)| {
+        serde_json::json!({"file_uuid": u, "file_name": n, "has_data": hd})
    }).collect();
    Ok(serde_json::json!({"found": true, "files": files}).to_string())
 }
@@ -197,17 +202,20 @@ async fn exec_find_file(pool: &sqlx::PgPool, args: &serde_json::Value) -> Result
 async fn exec_list_files(pool: &sqlx::PgPool, args: &serde_json::Value) -> Result<String, String> {
    let limit = args.get("limit").and_then(|v| v.as_i64()).unwrap_or(10);
    let videos = schema::table_name("videos");
-    let rows: Vec<(String, String)> = sqlx::query_as(&format!(
-        "SELECT file_uuid::text, file_name FROM {} ORDER BY created_at DESC LIMIT $1",
-        videos
+    let fd_table = schema::table_name("face_detections");
+    let rows: Vec<(String, String, bool)> = sqlx::query_as(&format!(
+        "SELECT v.file_uuid::text, v.file_name, \
+                (SELECT COUNT(*) FROM {} fd WHERE fd.file_uuid = v.file_uuid) > 0 AS has_data \
+         FROM {} v ORDER BY v.created_at DESC LIMIT $1",
+        fd_table, videos
    ))
    .bind(limit)
    .fetch_all(pool)
    .await
    .map_err(|e| e.to_string())?;

-    let files: Vec<serde_json::Value> = rows.into_iter().map(|(u, n)| {
-        serde_json::json!({"file_uuid": u, "file_name": n})
+    let files: Vec<serde_json::Value> = rows.into_iter().map(|(u, n, hd)| {
+        serde_json::json!({"file_uuid": u, "file_name": n, "has_data": hd})
    }).collect();
    Ok(serde_json::json!({"files": files}).to_string())
 }
--- a/src/api/five_w1h_agent_api.rs
+++ b/src/api/five_w1h_agent_api.rs
@@ -96,27 +96,11 @@ struct SceneSummaryResult {
 // ── LLM Endpoint ──

 fn llm_base_url() -> String {
-    let v = std::env::var("MOMENTRY_LLM_URL");
-    if v.is_ok() {
-        return v.unwrap();
-    }
-    let v = std::env::var("MOMENTRY_LLM_SUMMARY_URL");
-    if v.is_ok() {
-        return v.unwrap();
-    }
-    "http://localhost:8082/v1/chat/completions".to_string()
+    crate::core::config::llm::SUMMARY_URL.clone()
 }

 fn llm_model() -> String {
-    let v = std::env::var("MOMENTRY_LLM_MODEL");
-    if v.is_ok() {
-        return v.unwrap();
-    }
-    let v = std::env::var("MOMENTRY_LLM_SUMMARY_MODEL");
-    if v.is_ok() {
-        return v.unwrap();
-    }
-    "google_gemma-4-26B-A4B-it-Q5_K_M.gguf".to_string()
+    crate::core::config::llm::SUMMARY_MODEL.clone()
 }

 // ── Data Fetching ──
--- a/src/api/identity_api.rs
+++ b/src/api/identity_api.rs
@@ -668,7 +668,7 @@ pub struct RegisterResourceRequest {
 pub struct ResourceResponse {
    pub success: bool,
    pub message: String,
-    pub data: Option<ResourceItem>,
+    pub data: Option<Vec<ResourceItem>>,
 }

 #[derive(Debug, Serialize)]
@@ -677,6 +677,8 @@ pub struct ResourceItem {
    pub resource_type: String,
    pub category: String,
    pub capabilities: Option<serde_json::Value>,
+    pub config: Option<serde_json::Value>,
+    pub metadata: Option<serde_json::Value>,
    pub status: String,
    pub last_heartbeat: Option<chrono::DateTime<chrono::Utc>>,
 }
@@ -743,22 +745,24 @@ async fn list_resources(
        .await
        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;

-    let data: Vec<ResourceItem> = records
-        .into_iter()
-        .map(|r| ResourceItem {
-            resource_id: r.resource_id,
-            resource_type: r.resource_type,
-            category: r.category,
-            capabilities: r.capabilities,
-            status: r.status,
-            last_heartbeat: r.last_heartbeat,
-        })
-        .collect();
+let data: Vec<ResourceItem> = records
+    .into_iter()
+    .map(|r| ResourceItem {
+        resource_id: r.resource_id,
+        resource_type: r.resource_type,
+        category: r.category,
+        capabilities: r.capabilities,
+        config: r.config,
+        metadata: r.metadata,
+        status: r.status,
+        last_heartbeat: r.last_heartbeat,
+    })
+    .collect();

    Ok(Json(ResourceResponse {
        success: true,
        message: "Resources listed".to_string(),
-        data: None,
+        data: Some(data),
    }))
 }

--- a/src/api/server.rs
+++ b/src/api/server.rs
@@ -1,4 +1,7 @@
+use std::time::Duration;
+
 use axum::Router;
+use tokio::time::timeout;
 use tower_http::cors::{Any, CorsLayer};

 use crate::core::cache::{MongoCache, RedisCache};
@@ -30,7 +33,21 @@ pub async fn start_server(host: &str, port: u16) -> anyhow::Result<()> {
    health::init_server_state(host, port);

    let embedder = std::sync::Arc::new(Embedder::new("nomic-embed-text-v2-moe:latest".to_string()));
-    let mongo_cache = MongoCache::init().await?;
+
+    // MongoDB is ONLY a cache layer — if unavailable, the server continues
+    // with Redis cache alone. This keeps both 3002 and 3003 bootable
+    // without requiring MongoDB to be installed or running.
+    let mongo_cache = match timeout(Duration::from_secs(5), MongoCache::init()).await {
+        Ok(Ok(cache)) => cache,
+        Ok(Err(e)) => {
+            tracing::warn!("MongoDB cache unavailable (continuing without): {e}");
+            MongoCache::disabled().await
+        }
+        Err(_) => {
+            tracing::warn!("MongoDB init timed out (continuing without cache)");
+            MongoCache::disabled().await
+        }
+    };
    let redis_cache = RedisCache::new()?;
    let db = PostgresDb::init().await?;

--- a/src/core/cache/mongo_cache.rs
+++ b/src/core/cache/mongo_cache.rs
@@ -80,6 +80,25 @@ impl MongoCache {
        Ok(cache)
    }

+    /// Create a disabled cache instance — all ops are no-ops (is_enabled() = false).
+    /// Used when MongoDB is unavailable; Redis cache continues independently.
+    pub async fn disabled() -> Self {
+        let client = Client::with_uri_str("mongodb://localhost:27017")
+            .await
+            .expect("disabled mongo client (lazy — no actual connect)");
+        let db = client.database("disabled_cache");
+        Self {
+            client,
+            db: db.clone(),
+            collection: db.collection("disabled"),
+            settings: CacheSettings {
+                enabled: false,
+                ..Default::default()
+            },
+            initialized: Arc::new(RwLock::new(false)),
+        }
+    }
+
    async fn ensure_indexes(&self) -> Result<()> {
        let mut guard = self.initialized.write().await;
        if *guard {
--- a/src/core/config.rs
+++ b/src/core/config.rs
@@ -216,13 +216,47 @@ pub mod cache {
 pub mod llm {
    use super::*;

+    /// Chat / function-calling LLM endpoint (agents/search, translation, etc.)
+    /// Default: http://127.0.0.1:8082/v1/chat/completions
+    pub static CHAT_URL: Lazy<String> = Lazy::new(|| {
+        env::var("MOMENTRY_LLM_CHAT_URL")
+            .or_else(|_| env::var("MOMENTRY_LLM_SUMMARY_URL"))
+            .or_else(|_| env::var("MOMENTRY_LLM_URL"))
+            .unwrap_or_else(|_| "http://127.0.0.1:8082/v1/chat/completions".to_string())
+    });
+
+    pub static CHAT_MODEL: Lazy<String> = Lazy::new(|| {
+        env::var("MOMENTRY_LLM_CHAT_MODEL")
+            .or_else(|_| env::var("MOMENTRY_LLM_SUMMARY_MODEL"))
+            .or_else(|_| env::var("MOMENTRY_LLM_MODEL"))
+            .unwrap_or_else(|_| "google_gemma-4-26B-A4B-it-Q5_K_M.gguf".to_string())
+    });
+
+    /// Vision LLM endpoint (frame analysis, OCR). Can be same as CHAT_URL or different.
+    /// Default: falls back to CHAT_URL
+    pub static VISION_URL: Lazy<String> = Lazy::new(|| {
+        env::var("MOMENTRY_LLM_VISION_URL")
+            .unwrap_or_else(|_| CHAT_URL.clone())
+    });
+
+    pub static VISION_MODEL: Lazy<String> = Lazy::new(|| {
+        env::var("MOMENTRY_LLM_VISION_MODEL")
+            .unwrap_or_else(|_| CHAT_MODEL.clone())
+    });
+
+    /// Text summary LLM endpoint (5W1H, story). Can be same as CHAT_URL or different.
    pub static SUMMARY_URL: Lazy<String> = Lazy::new(|| {
        env::var("MOMENTRY_LLM_SUMMARY_URL")
-            .unwrap_or_else(|_| "http://127.0.0.1:8081/v1/chat/completions".to_string())
+            .ok()
+            .or_else(|| Some(CHAT_URL.clone()))
+            .unwrap()
    });

    pub static SUMMARY_MODEL: Lazy<String> = Lazy::new(|| {
-        env::var("MOMENTRY_LLM_SUMMARY_MODEL").unwrap_or_else(|_| "gemma4".to_string())
+        env::var("MOMENTRY_LLM_SUMMARY_MODEL")
+            .ok()
+            .or_else(|| Some(CHAT_MODEL.clone()))
+            .unwrap()
    });

    pub static SUMMARY_TIMEOUT_SECS: Lazy<u64> = Lazy::new(|| {
@@ -237,6 +271,13 @@ pub mod llm {
            .map(|v| v == "true" || v == "1")
            .unwrap_or(true)
    });
+
+    pub static CHAT_TIMEOUT_SECS: Lazy<u64> = Lazy::new(|| {
+        env::var("MOMENTRY_LLM_CHAT_TIMEOUT")
+            .unwrap_or_else(|_| "120".to_string())
+            .parse()
+            .unwrap_or(120)
+    });
 }

 pub static SFTPGO_BASE_URL: Lazy<String> = Lazy::new(|| {
--- a/src/core/llm/function_calling.rs
+++ b/src/core/llm/function_calling.rs
@@ -1,6 +1,8 @@
 use serde::{Deserialize, Serialize};
 use serde_json::{json, Value};

+use crate::core::config;
+
 /// A tool/function definition for Gemma4 function calling
 #[derive(Debug, Clone, Serialize)]
 pub struct ToolDef {
@@ -75,18 +77,24 @@ pub enum LlmResponse {
    ToolCalls(Vec<ToolCall>),
 }

-/// Get the LLM chat URL with fallback chain
+/// Get the LLM chat URL from centralized config
 pub fn llm_chat_url() -> String {
-    std::env::var("MOMENTRY_LLM_URL")
-        .or_else(|_| std::env::var("MOMENTRY_LLM_SUMMARY_URL"))
-        .unwrap_or_else(|_| "http://localhost:8082/v1/chat/completions".to_string())
+    config::llm::CHAT_URL.clone()
 }

-/// Get the LLM model name
+/// Get the LLM model name from centralized config
 pub fn llm_model() -> String {
-    std::env::var("MOMENTRY_LLM_MODEL")
-        .or_else(|_| std::env::var("MOMENTRY_LLM_SUMMARY_MODEL"))
-        .unwrap_or_else(|_| "google_gemma-4-26B-A4B-it-Q5_K_M.gguf".to_string())
+    config::llm::CHAT_MODEL.clone()
+}
+
+/// Get the vision LLM URL
+pub fn llm_vision_url() -> String {
+    config::llm::VISION_URL.clone()
+}
+
+/// Get the vision LLM model name
+pub fn llm_vision_model() -> String {
+    config::llm::VISION_MODEL.clone()
 }

 /// Build a tool definition JSON for function calling
@@ -113,7 +121,9 @@ pub async fn call_llm(
    timeout_secs: u64,
 ) -> anyhow::Result<LlmResponse> {
    let client = reqwest::Client::builder()
-        .timeout(std::time::Duration::from_secs(timeout_secs))
+        .timeout(std::time::Duration::from_secs(
+            if timeout_secs > 0 { timeout_secs } else { *config::llm::CHAT_TIMEOUT_SECS },
+        ))
        .build()?;

    let req = ChatRequest {
Author	SHA1	Message	Date
M5Max128	1c30af9557	fix: correct service paths, nohup removal, MongoDB graceful fallback, add MariaDB + Caddy to startup - Fix Qdrant binary path (services/ -> momentry_resources/bin/) - Fix LLM binary/model paths (llama/ -> momentry_resources/llama/, models/ -> models/llm/) - Fix PostgreSQL data path (pgsql/data -> momentry/var/postgresql) - Remove nohup (fails in LaunchDaemon environment) - Add MongoDB graceful fallback with 5s timeout in server.rs - Add MariaDB + Caddy steps to startup script for WordPress - Revert all unrelated changes	2026-05-23 01:46:23 +08:00
M5Max128	3ccdf403b6	feat: add Ollama to verified sources (Gitea repo + manifest + build from source)	2026-05-22 17:20:14 +08:00
Accusys	c09268f3d3	docs: add go(golang) and ollama verification reports	2026-05-22 16:58:08 +08:00
Accusys	84a2f71e30	docs: add verification_doc links to service sources manifest	2026-05-22 16:57:45 +08:00
Accusys	9b32d1fed4	docs: add Gitea repo URLs to service sources manifest	2026-05-22 16:45:55 +08:00
Accusys	3ef2e6e150	docs: add service sources manifest (replace src/ directory)	2026-05-22 16:38:58 +08:00
Accusys	c4e30e4234	fix: list_resources returns data (config+metadata); register source code resource	2026-05-22 16:01:33 +08:00
Accusys	bd82028f34	refactor: unified LLM config - CHAT_URL/VISION_URL/SUMMARY_URL with env var overrides	2026-05-22 15:47:17 +08:00
Accusys	a78b5bc12b	docs: add agents/search endpoint to 12_agent.md	2026-05-22 12:26:11 +08:00
Accusys	2d008b75bf	fix: find_file/list_files include has_data flag for video data availability	2026-05-22 12:22:35 +08:00