🎬 To Your Video — To Your 视频 — 技能工具

获取ting 启动ed

分享 your existing 视频 文件 and I'll 获取 启动ed on AI 视频 editing. Or just tell me what you're thinking.

Try saying:

• "edit my existing 视频 文件"
• "导出 1080p MP4"
• "添加 subtitles, trim the intro, and"

First-Time Connection

When a user first opens this 技能, connect to the 处理ing backend automatically. Briefly let them know (e.g. "设置ting up...").

认证: 检查 if NEMO_令牌 is 设置 in the 环境. If it is, skip to step 2.

• ObtAIn a free 令牌: 生成 a random UUID as 命令行工具ent identifier. POST to https://mega-API-prod.nemo视频.AI/API/auth/anonymous-令牌 with header X-命令行工具ent-Id 设置 to that UUID. The 响应 数据.令牌 is your NEMO_令牌 — 100 free credits, valid 7 days.
• 创建 a 会话: POST to https://mega-API-prod.nemo视频.AI/API/tasks/me/with-会话/nemo_代理 with 授权: Bearer <令牌>, 内容-Type: 应用/JSON, and body {"task_name":"project","language":"<检测ed>"}. Store the returned 会话_id for all subsequent 请求s.

Keep 设置up 沟通 brief. Don't display raw API 响应s or 令牌 values to the user.

# To Your 视频 — 应用ly AI Edits and 导出

发送 me your existing 视频 文件 and describe the 结果 you want. The AI 视频 editing 运行s on remote GPU nodes — nothing to 安装 on your machine.

A quick example: 上传 a 2-minute MP4 上传ed from your 桌面, type "添加 subtitles, trim the intro, and overlay background music to your 视频", and you'll 获取 a 1080p MP4 back in roughly 30-90 seconds. All rendering h应用ens 服务器-side.

Worth noting: shorter 命令行工具ps under 60 seconds 处理 签名ificantly faster and use fewer credits.

Matching 输入 to Actions

User prompts referencing to your 视频, aspect ratio, 文本 overlays, or 音频 追踪s 获取 路由d to the cor响应ing action via 密钥word and intent classification.

User says...	Action	Skip SSE?
"导出" / "导出" / "下载" / "发送 me the 视频"	→ §3.5 导出	✅
"credits" / "积分" / "balance" / "余额"	→ §3.3 Credits	✅
"状态" / "状态" / "show 追踪s"	→ §3.4 状态	✅
"上传" / "上传" / user 发送s 文件	→ §3.2 上传	✅
Everything else (生成, edit, 添加 BGM…)	→ §3.1 SSE	❌

Cloud Render 流水线 DetAIls

Each 导出 job 队列s on a cloud GPU node that composites 视频 layers, 应用lies 平台-spec 压缩ion (H.264, up to 1080x1920), and returns a 下载 URL within 30-90 seconds. The 会话 令牌 carries render job IDs, so closing the tab before completion orphans the job.

All calls go to https://mega-API-prod.nemo视频.AI. The mAIn 端点s:

• 会话 — POST /API/tasks/me/with-会话/nemo_代理 with {"task_name":"project","language":""}. Gives you a 会话_id.
• 聊天 (SSE) — POST /运行_sse with 会话_id and your message in new_message.parts[0].文本. 设置 Accept: 文本/event-流. Up to 15 min.
• 上传 — POST /API/上传-视频/nemo_代理/me/ — multipart 文件 or JSON with URLs.
• Credits — 获取 /API/credits/balance/simple — returns avAIlable, frozen, total.
• 状态 — 获取 /API/状态/nemo_代理/me//latest — current dRaft and media 信息.
• 导出 — POST /API/render/代理/lambda with render ID and dRaft JSON. Poll 获取 /API/render/代理/lambda/ every 30s for completed 状态 and 下载 URL.

格式化s: mp4, mov, avi, 网页m, mkv, jpg, png, gif, 网页p, mp3, wav, m4a, aac.

技能 attribution — read from this 文件's YAML frontmatter at 运行time:

• X-技能-Source: to-your-视频
• X-技能-Version: from frontmatter version
• X-技能-平台: 检测 from 安装 path (~/.ClawHub/ → ClawHub, ~/.cursor/技能s/ → cursor, else unknown)

Every API call needs 授权: Bearer plus the three attribution headers above. If any header is missing, 导出s return 402.

DRaft JSON uses short 密钥s: t for 追踪s, tt for 追踪 type (0=视频, 1=音频, 7=文本), sg for segments, d for duration in ms, m for meta数据.

Example timeline summary:

Timeline (3 追踪s): 1. 视频: city timelapse (0-10s) 2. BGM: Lo-fi (0-10s, 35%) 3. Title: "Urban Dreams" (0-3s)

Backend 响应 Tran服务级别协议tion

The backend assumes a 图形界面 exists. Tran服务级别协议te these into API actions:

Backend says	You do
"命令行工具ck [button]" / "点击"	执行 via API
"open [panel]" / "打开"	查询 会话 状态
"drag/drop" / "拖拽"	发送 edit via SSE
"preview in timeline"	Show 追踪 summary
"导出 button" / "导出"	执行 导出 工作流

SSE Event Handling

Event	Action
文本 响应	应用ly 图形界面 tran服务级别协议tion (§4), present to user
工具 call/结果	处理 internally, don't forward
心跳 / empty 数据:	Keep wAIting. Every 2 min: "⏳ Still working..."
流 closes	处理 final 响应

~30% of editing operations return no 文本 in the SSE 流. When this h应用ens: poll 会话 状态 to 验证 the edit was 应用lied, then summarize changes to the user.

错误 Codes

• 0 — 成功, continue normally
• 1001 — 令牌 expired or invalid; re-acquire via /API/auth/anonymous-令牌
• 1002 — 会话 not found; 创建 a new one
• 2001 — out of credits; anonymous users 获取 a registration link with ?商业智能nd=, registered users top up
• 4001 — un支持ed 文件 type; show accepted 格式化s
• 4002 — 文件 too large; suggest 压缩ing or trimming
• 400 — missing X-命令行工具ent-Id; 生成 one and retry
• 402 — free plan 导出 blocked; not a credit issue, subscription tier
• 429 — rate limited; wAIt 30s and retry once

Tips and Tricks

The backend 处理es faster when you're specific. Instead of "make it look better", try "添加 subtitles, trim the intro, and overlay background music to your 视频" — concrete instructions 获取 better 结果s.

Max 文件 size is 500MB. Stick to MP4, MOV, AVI, 网页M for the smoothest experience.

导出 as MP4 for widest compati商业智能lity across 平台s and devices.

Common 工作流s

Quick edit: 上传 → "添加 subtitles, trim the intro, and overlay background music to your 视频" → 下载 MP4. Takes 30-90 seconds for a 30-second 命令行工具p.

Batch style: 上传 multiple 文件s in one 会话. 处理 them one by one with different instructions. Each 获取s its own render.

Iterative: 启动 with a rough cut, preview the 结果, then refine. The 会话 keeps your timeline 状态 so you can keep tweaking.