adobe-edit-quick-cut

Files (4)

• SKILL.md
• openai.yaml · agents/
• icon-large.svg · assets/
• icon-small.svg · assets/

SKILL.md (first lines)

# Adobe Edit Quick Cut

Produces 3 AI-edited sizzle reel variations from a source video, all at the same duration and
style — giving the user options to pick from.

> **Surface note:** The default flow below uses Adobe's MCP App widgets — the `asset_add_file` file picker (Step 2) and the side-by-side `asset_preview_file` preview (Step 7). Follow it as written. Only if a widget tool is **not available on this surface** (e.g. Codex), use the *No-widget fallback* attached to that step.

---

## Tool Reference

| Step | Tool | Notes |
|------|------|-------|
| Upload source video | `asset_add_file` | File picker; returns the CC asset ID required by Quick Cut |
| Stage source video *(no-widget fallback)* | `asset_initialize_file_upload` + `asset_finalize_file_upload` | Only when `asset_add_file` is unavailable — stage a local video to CC; extract `assetId` from the finalize response |
| Run Quick Cut variations | `video_create_quick_cut` | Fire 3 in parallel; same duration and style prompt |
| Poll job status | `quickCutPoll` | Repeat until all 3 return `completed` |
| Preview variations | `asset_preview_file` | Renders all 3 side-by-side for selection *(no-widget fallback: present the 3 URLs directly)* |
| Resize re-uploaded output | `video_resize` | Workaround only — Quick Cut output must be re-uploaded first |

---

## Workflow

### Step 0 — Initialize Adobe Tools

Call `adobe_mandatory_init` first. This returns file handling rules and tool routing guidance required for the rest of the workflow.

```json
{ "skill_name": "adobe-edit-quick-cut", "skill_version": "2.1.0" }
```

---

### Step 1 — Entitlement Check

Now that `adobe_mandatory_init` confirmed that the "Adobe for creativity" connector is live, check which tools are available through the "Adobe for creativity" connector by cross checking against the Tool Reference table above.

This also tells you which widgets this surface supports: if `asset_add_file` and `asset_preview_file` are available, follow the default flow (Steps 2 and 7 as written). If one is not available (e.g. Codex), use that step's *No-widget fallback*. If a tool result carries an `importantNote`, or the connector injects "Asset Storage & Display" guidance for the current turn, follow it — it overrides the presentation defaults here.

---

### Step 2 — Open the File Picker

Open the picker immediately with this message:

> *"Let's create a punchy sizzle reel from your video. Start by selecting your file:"*

```javascript
asset_add_file()
```

Once the user selects a file, extract `assetId` (the CC asset ID) from the widget context.

> `video_create_quick_cut` requires a CC asset ID (`assetId`), not `presignedAssetUrl`.

**No-widget fallback** *(only if `asset_add_file` is unavailable on this surface, e.g. Codex)* — don't ask the user to pick; get the `assetId` from where the file is. Staging a local file requires egress — check egress status from `adobe_mandatory_init` first; if egress is disabled and no picker is available on this surface, tell the user staging isn't possible here. Otherwise:

| Source | Action |
|--------|--------|
| File at a local path (e.g. `/mnt/user-data/uploads/…`) | Stage programmatically: get file size and MIME type, call `asset_initialize_file_upload({ path: "<filename>", media_type: "<mime>" })`, PUT the bytes to the returned upload URL, then `asset_finalize_file_upload({ filename: "<filename>", transfer_document: <from initialize response> })`. Extract the `assetId` from the finalize response — this is what `video_create_quick_cut` needs. |
| File already in Creative Cloud | Reference it directly by its CC `assetId`. |

---

### Step 3 — Confirm Upload

Once the file is selected, confirm with:

> *"Got it — [filename] is ready. Now let's set up your cut."*

Then immediately present the Q&A form below.

---

### Step 4 — Q&A Form (via AskUserQuestion)

Wait for the user's answers before proceeding; present the questions via `AskUserQuestion` (not plain text) so the user gets tappable buttons.

> **No-widget fallback** *(only if `AskUserQuestion` is unavailable, e.g. Codex)* — ask the same labeled options as a plain-text message and wait for the user's typed reply. The option set is identical.

```javascript
AskUserQuestion({
  questions: [
    {
      header: "Cut Length",
      question: "What kind of cut would you like? (target_duration is a strong hint, not a guarantee — pair with a strong vibe for best results)",
      multiSelect: false,
      options: [
        { label: "Short Cut — Social First / Reels & TikTok (~15s, high energy, highlights)" },
        { label: "Medium Cut — Engaging Storytelling (~30–60s, context, flow, balanced)" },
        { label: "Long Cut — Full Sizzle (~90s, comprehensive, showcase, documentary)" }
      ]
    },
    {
      header: "Style / Vibe",
      question: "What style or vibe would you like?",
      multiSelect: false,
      options: [
        { label: "Action & Energy" },
        { label: "Key Talking Moments" },
        { label: "Cinematic & Dramatic" },
        { label: "No Preference" }
      ]
    }
  ]
})
```

Wait for the user's selections before proceeding to Step 5.

---

### Step 5 — Acknowledge and Run

Once the user answers, respond with:

> *"Got it — [cut type], [style] vibe. Creating 3 variations at that length — let me preview them for you."*

From xixu-me/scps, Apache-2.0.

More from xixu-me/scps