Connect or check the local LEOR account connection. If LEOR is already connected, this returns connected=true and no authorizationUrl; do not treat that as failure. If authorization is missing or forceReconnect=true, it returns a fresh authorizationUrl. Do not inspect local .env, MCP config, source files, token values, or backend code.

该工具无需必填参数，直接调用即可

Free tool. Check LEOR Agent Gateway version, playbook version, and latest update notice. Use once when a session starts, after reconnecting LEOR, or when behavior seems stale. The gateway suppresses repeated reminders for the same noticeId, so do not manually repeat the same update notice.

该工具无需必填参数，直接调用即可

Free read-only tool. Check the currently authorized LEOR account status and credit balance. Use when the user asks how many credits are left, whether their LEOR account is connected, or whether they have enough credits before generation. It does not spend credits and does not expose tokens or private credentials.

该工具无需必填参数，直接调用即可

Free preflight checker for LEOR credit-consuming generation tools. Use when the user wants AI image generation but the brief may be incomplete or ambiguous. It never spends credits; it returns readyToGenerate, missingFields, questionsToAsk, recommendedQuestions, reminders, and which paid tool can be called. For first-round booth generation, do not rely on defaults: collect the LEOR main left-sidebar brief, ask logo/reference decisions, summarize, and wait for explicit user confirmation before spending credits.

该工具无需必填参数，直接调用即可

Search public approved LEOR inspiration gallery images. Use before generation when the user is vague, asks for references/cases, says 找/搜/看看/灵感广场/案例/参考图, or wants to use an existing LEOR gallery image as visual direction. A follow-up color such as 橙色方案 after a search request still means refine the gallery search; do not switch to generation. Use query for free-text words such as 建材展, 瓷砖, 橙色, or 色彩鲜艳; use industry only for exact LEOR category labels. If one search is empty, try related words before telling the user no close reference was found.

该工具无需必填参数，直接调用即可

Get one LEOR inspiration gallery item with safe preview metadata after the user selects a reference. Search results return galleryItemId; pass that value here as galleryItemId or itemId. Use previewUrl directly for conversation display when the host supports image URLs; it is a watermarked inline preview, not the raw original. Use downloadUrl only when the host cannot display image URLs or the user explicitly asks for a local file; downloadUrl is also watermarked. Use galleryItemId for generation instead of downloading or exposing raw gallery image URLs.

该工具无需必填参数，直接调用即可

Credit-consuming: generate exactly 1 booth design image from a user brief in Agent/MCP. Use only when the user clearly wants a new generated booth image, not when they asked to find/search/browse inspiration references. Do not call this for color-only or style-only phrases such as 橙色方案. New booth design corresponds to the real LEOR website left sidebar: 客户与品牌(clientName/logo/referenceImage/industry), 空间与布局(width/depth/height/openingLayout), 设计方向(style/colors/boothShape/additionalNotes), 输出设置(resolution/aspectRatio/renderModel/count/grid). Before the first paid booth generation, give the user one copyable checklist split into Required and Optional-but-ask fields. Required: brand/company name, display product, width/depth/height, front/back/left/right wall-opening layout, visual style, colors, booth shape, materials/functions/details, render model, resolution, aspect ratio. Optional-but-ask: logo and reference image; assets are optional but the yes/no decision is required. If the user attached local images in the agent conversation, pass them with logoImageBase64/referenceImageBase64. These are transport fields only: LEOR stores the images in the authorized user project assets and uses generated asset URLs internally. Do not tell the user LEOR only accepts image URLs, and do not ask the user to convert files to base64. Agent/MCP never generates multiple images in one call. If the user asks for multiple directions, explain that LEOR will generate them one by one, each charged separately, and ask whether to start the first image. Then summarize the brief and wait for explicit user confirmation. If required brief fields are missing the tool returns needsMoreBrief and will not spend credits. LEOR backend handles prompt assembly and asset storage without exposing prompts.

必填参数：prompt

Refine an existing booth image by applying a user instruction such as 更科技, 更高端, 加洽谈区, 换材质, 加大屏, 强化门头. Use after the user selects a generated/reference image.

必填参数：prompt

One-click LEOR node tool. Generate a 3x3 booth touring/storytelling storyboard from the current booth image. Use exact nodeToolMode storyboard for 展位分镜, 游览动线, 多角度展示, or client proposal visuals.

该工具无需必填参数，直接调用即可

One-click LEOR node tool. Generate a booth color/material board from the current booth image. Use exact nodeToolMode material_board for 材质看板, 色彩材质方案, construction direction, or proposal material direction.

该工具无需必填参数，直接调用即可

One-click LEOR node tool. 万物成展 is for early-stage creative ideation: turn an arbitrary visual source, usually not an existing booth, into an exhibition booth concept by translating visual DNA such as color, shape, structure, material, lighting, and mood. Good sources include animals, fish, fruit, vegetables, logos, product photos, packaging, color swatches, textures, objects, and brand visuals. Do not use this as the normal tool for evolving an existing booth; use evolve_booth_image for booth changes, and use generate_booth_directions with galleryItemId/sourceImageUrl when a LEOR gallery booth is only a reference. Use exact nodeToolMode everything_to_booth for 万物成展.

该工具无需必填参数，直接调用即可

Credit-consuming one-step image task. Use when the user sends an existing booth image and clearly asks to modify it. LEOR stores the source image in the user agent project, creates an output node, runs evolve, and returns a task id for polling. Requires one JPG/PNG/WEBP image up to 25MB and a clear change instruction. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends a booth image and asks for 展位分镜, 游览动线, multi-angle storyboard, or proposal walkthrough. LEOR stores the source image in the user agent project, creates an output node, runs storyboard, and returns a task id for polling. Accepts one JPG/PNG/WEBP image up to 25MB. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends a booth image and asks for 材质看板, color/material board, or proposal material direction. LEOR stores the source image in the user agent project, creates an output node, runs material_board, and returns a task id for polling. Accepts one JPG/PNG/WEBP image up to 25MB. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends an image and clearly wants to use that same image for 万物成展. Do not expose a separate upload step to the user. LEOR stores the source image in the user agent project, creates an output node, runs everything_to_booth, and returns a task id for polling. Accepts only one JPG/PNG/WEBP image up to 25MB, matching the main LEOR upload rules. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends a booth image and asks for 潮流海报, KV-like booth promotion, or energetic proposal poster. LEOR stores the source image in the user agent project, creates an output node, runs trend_poster, and returns a task id for polling. Accepts one JPG/PNG/WEBP image up to 25MB. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends a booth image and asks for 简约海报, PPT cover, or restrained proposal cover. LEOR stores the source image in the user agent project, creates an output node, runs minimal_poster, and returns a task id for polling. Accepts one JPG/PNG/WEBP image up to 25MB. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming one-step image task. Use when the user sends a booth image and asks for 玻璃质感, premium archive cover, or restrained brand atmosphere poster. LEOR stores the source image in the user agent project, creates an output node, runs glass_poster, and returns a task id for polling. Accepts one JPG/PNG/WEBP image up to 25MB. Not a generic upload or bulk image hosting tool.

必填参数：imageBase64

Credit-consuming node tool. Generate one of LEOR’s 14 extra preset camera angles from an existing booth image. User-facing name: 旋转视角. Recognize older/synonym phrases such as 转换视角, 生成角度图, 换个角度, 多角度效果图, but reply to users with 旋转视角. Requires sourceImageUrl from a current LEOR image or visible image URL. Valid anglePreset values: left90_low, left90_eye, left90_top, left45_low, left45_eye, left45_top, front_low, front_top, right45_low, right45_eye, right45_top, right90_low, right90_eye, right90_top. Use right45_eye by default when the user only says 旋转视角. Poll with get_generation_result.

该工具无需必填参数，直接调用即可

One-click LEOR node tool. Generate a high-impact commercial proposal poster from the current booth image. Use exact nodeToolMode trend_poster for 潮流海报, KV-like booth promotion, social sharing, or energetic proposal visuals.

该工具无需必填参数，直接调用即可

One-click LEOR node tool. Generate a restrained high-end minimal proposal cover from the current booth image. Use exact nodeToolMode minimal_poster for 简约海报, PPT cover, client proposal title image, or quiet premium presentation visuals.

该工具无需必填参数，直接调用即可

One-click LEOR node tool. Generate an embossed/frosted-glass portfolio-style cover from the current booth image. Use exact nodeToolMode glass_poster for 玻璃质感, premium archive cover, or restrained brand atmosphere image.

该工具无需必填参数，直接调用即可

Fetch task status and final result from LEOR by task id. Poll after generation until status is done or error, then return fullResUrl/image URLs and project/item ids when available. The progress number is an approximate LEOR stage indicator, not the real model-side completion percentage; it may jump between preparation, generation, saving, and done. In user-facing replies, prefer the message/status stage over exact percentages unless the user asks for technical status.

必填参数：taskId