{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_f7nvtgsnqmg4","handle":"image-to-design-mcp","url":"https://wellknown.network/agents/image-to-design-mcp","links":{"self":"https://wellknown.network/agents/image-to-design-mcp/record.json","html":"https://wellknown.network/agents/image-to-design-mcp","markdown":"https://wellknown.network/agents/image-to-design-mcp/record.md","api":"https://wellknown.network/api/v1/agents/image-to-design-mcp","status":"https://wellknown.network/api/v1/agents/image-to-design-mcp/status","claim":"https://wellknown.network/agents/image-to-design-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/image-to-design-mcp/claim.json","badge":"https://wellknown.network/agents/image-to-design-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/image-to-design-mcp/history","tools":"https://wellknown.network/api/v1/agents/image-to-design-mcp/tools"},"ard":{"identifier":"urn:air::server:image-to-design-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"image-to-design-mcp","summary":"MCP server for image-to-design workflow automation","description":"# gauss-mcp\n\n基于 `uv` 管理的 Python MCP server，用于把本地图片导入为方案，并通过 job 协议提供可轮询的进度与结果。\n\n## 当前 MVP 范围\n\n- 输入本地图片路径\n- 上传图片\n- 判断普通图/全景图\n- 普通图转全景图\n- 可选清家具\n- 全景图转点云\n- 空间调整\n- 导入方案\n- 返回 design link\n\n## 核心设计\n\n- 普通 tool 调用保持短切片，不等待整条 job 结束\n- 对外主流程暴露 7 个 job tools：\n  - `gauss_start_image_job`\n  - `gauss_update_job_inputs`\n  - `gauss_wait_current_stage`\n  - `gauss_poll_job`\n  - `gauss_get_job`\n  - `gauss_resume_job`\n  - `gauss_list_jobs`\n- `gauss_wait_current_stage` 是默认等待手段：服务端内部执行 `poll + wait`，只等待“当前阶段结束”或一个 `timeout_sec` 切片，默认 60 秒；超时返回 `status=in_progress`\n- `gauss_poll_job` 保留给 debug 或细粒度观察，不再要求 agent 自己 Bash `sleep`\n- 对于用户未提前给出的阶段参数，可以在 `gauss_start_image_job` 时显式传 `null`，等进入对应阶段前再通过 `gauss_update_job_inputs` 补齐\n- `clear_furniture` 只在 `pano_clear` 阶段生效，`level_height_mm` 只在 `gauss_import` 阶段生效\n- 当 job 进入缺参数阶段时，紧凑 `job.required_user_input` 会给出结构化提问信号，`default_allowed=false` 表示调用方必须先问用户，不能自行默认\n- `gauss_wait_current_stage` 在这种场景下会返回 `status=needs_user_input`，而不是继续自动等待\n- 通过阶段事件与阶段摘要给 skill 提供进度展示依据，而不是 fake percent\n- 紧凑 `job` 返回会在已有可展示全景图时携带 `pano_url`，供 skill 直接展示图片反馈\n- Claude Code skill 不会在任务一开始统一追问参数；只有在 `pano_clear` / `gauss_import` 开始前才会询问对应设置\n- 长耗时阶段按 15-30 秒节奏推进；同步 wait tool 会复用阶段轮询间隔，减少 transcript 噪音\n- `gauss_import` 完成后会立即产出 `design_url`，用户可以先打开方案预览；`gauss_show_info` 继续补齐 `splat_url` / `drc_url`\n- job 状态落盘到 `IMAGE_TO_DESIGN_JOB_STATE_DIR`，支持中断后继续查询或恢复\n\n## 阶段顺序\n\n1. `upload`\n2. `image_type`\n3. `img2pano`\n4. `pano_clear`\n5. `pano2pointcloud`\n6. `spatial_tune`\n7. `gauss_import`\n8. `gauss_show_info`\n\n## 开发\n\n```bash\nuv sync --dev\nuv run pytest\nuv run ruff check src tests\n```\n\n## 运行 MCP Server\n\n```bash\nuv run gauss-mcp\n```\n\n## 本地 smoke test\n\n默认使用同步 wait tool：\n\n```bash\nuv run python scripts/e2e_image_to_plan.py /absolute/path/to/image.png\n```\n\n可选参数：\n\n```bash\nuv run python scripts/e2e_image_to_plan.py /absolute/path/to/image.png \\\n  --floor-plan-name demo-plan \\\n  --level-height-mm 2800 \\\n  --no-clear-furniture \\\n  --wait-timeout-sec 60\n```\n\n如需切回细粒度 poll 调试：\n\n```bash\nuv run python s…","publisher":null,"homepage":null,"repository":null,"version":"0.1.6","license":null,"protocols":["mcp"],"tags":["mcp"],"pricing":null,"endpoints":[{"url":"pypi:image-to-design-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","summary":"pypi","version":"pypi","description":"pypi"}},"derived":{"capabilities":[{"slug":"media.3d","name":"3D & Design","confidence":1,"provenance":"derived"}],"categories":["media"],"language":"fr"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"image-to-design-mcp","registry":"pypi","observedAt":"2026-11-03T09:25:39.458Z","publishedAt":"2026-03-31T08:21:46.768040Z","latestVersion":"0.1.6"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"image-to-design-mcp","url":"https://pypi.org/project/image-to-design-mcp/","firstSeenAt":"2026-09-09T21:25:35.336Z","fetchedAt":"2026-11-03T09:24:24.676Z","normalizedAt":"2026-11-03T09:24:24.676Z"}]},"firstSeenAt":"2026-09-09T21:25:35.336Z","updatedAt":"2026-11-03T09:25:39.458Z"}