Agent Skill

# DeepManim API Skill

You can use the DeepManim API to generate, improve, and narrate animated explainer videos from text prompts.

## Base URL

```
https://api.deepmanim.com/api/v1
```

## Authentication

All requests require an API key in the Authorization header:

```
Authorization: Bearer dm_k_YOUR_API_KEY
```

## Workflow

The typical workflow is:

1. **Generate** a video from a text prompt via `POST /generate`
2. **Poll** the returned job via `GET /jobs/{job_id}` until `status` is `"completed"`
3. **Improve** the video with follow-up instructions via `POST /improve` (optional, repeatable)
4. **Add audio** narration via `POST /audio` after you're happy with the visuals
5. **Improve narration** via `POST /improve-narration` if needed

Generate and improve calls start at 1.5 credits and vary by preset. Audio and improve-narration calls cost 1 credit. Reading data is free.

## Endpoints

### POST /generate
Generate a new video from a text prompt. Audio narration is included by default.

Request body:
```json
{
  "message": "Explain the Pythagorean theorem",
  "session_id": null
}
```
- `message` (required): The prompt describing what to animate.
- `session_id` (optional): Pass an existing session ID to continue a conversation.

Response:
```json
{
  "job_id": "abc-123",
  "session_id": "def-456",
  "status": "pending"
}
```

### POST /improve
Improve or modify an existing video. Does NOT include audio — add it separately via /audio after all modifications.

Request body:
```json
{
  "session_id": "def-456",
  "message": "Make the background darker and slow down the animation"
}
```

Response:
```json
{
  "job_id": "ghi-789",
  "session_id": "def-456",
  "status": "pending"
}
```

### POST /audio
Add narration audio to an existing video.

Request body:
```json
{
  "session_id": "def-456",
  "message_id": "msg-123",
  "high_quality": false
}
```

### POST /improve-narration
Improve existing narration on a video.

Request body:
```json
{
  "session_id": "def-456",
  "message_id": "msg-123",
  "high_quality": true
}
```

### GET /jobs/{job_id}
Poll job status. The `phase` field indicates progress:
`brainstorming` → `crafting_vision` → `assembling_pieces` → `polishing` → `done`

Response when completed:
```json
{
  "job_id": "abc-123",
  "status": "completed",
  "phase": "done",
  "estimated_time_remaining_seconds": 0,
  "result": {
    "video_url": "https://...",
    "description": "An animation of the Pythagorean theorem...",
    "session_id": "def-456",
    "message_id": "msg-789",
    "duration_seconds": 24.6,
    "has_audio": true,
    "follow_up": "What are some real-world applications?"
  }
}
```

Possible `status` values: `pending`, `processing`, `completed`, `failed`, `cancelled`.

### POST /jobs/{job_id}/cancel
Cancel a running or pending job.

### GET /sessions
List all your sessions.

### GET /sessions/{session_id}
Get a session with full message history.

### DELETE /sessions/{session_id}
Delete a session and all its messages.

### GET /sessions/{session_id}/jobs
List jobs for a session. Optional query param: `?status=completed`

### GET /messages/{message_id}
Get a single message by ID.

### GET /credits
Get credit balance: `balance`, `total_used`, `total_purchased`, `plan`.

### GET /me
Get current user info: `id`, `email`, `display_name`.

## Error Codes

- `401` — Missing or invalid API key
- `402` — Insufficient credits
- `404` — Resource not found
- `400` — Invalid request

## Polling Strategy

Jobs typically take 60–120 seconds. Poll `GET /jobs/{job_id}` every 3–5 seconds. Use the `estimated_time_remaining_seconds` field to adjust polling frequency. Stop polling when `status` is `completed`, `failed`, or `cancelled`.

## Example: Full Workflow

```python
import requests, time

API_KEY = "dm_k_YOUR_KEY"
BASE = "https://api.deepmanim.com/api/v1"
headers = {"Authorization": f"Bearer {API_KEY}"}

# 1. Generate
r = requests.post(f"{BASE}/generate", headers=headers, json={"message": "Explain gravity"})
job_id = r.json()["job_id"]
session_id = r.json()["session_id"]

# 2. Poll until done
while True:
    job = requests.get(f"{BASE}/jobs/{job_id}", headers=headers).json()
    if job["status"] in ("completed", "failed", "cancelled"):
        break
    time.sleep(4)

# 3. Get the result
video_url = job["result"]["video_url"]
message_id = job["result"]["message_id"]

# 4. Optionally improve
r = requests.post(f"{BASE}/improve", headers=headers, json={
    "session_id": session_id,
    "message": "Add more color and make it slower"
})
# Poll the new job_id the same way...

# 5. Add audio after improvements
r = requests.post(f"{BASE}/audio", headers=headers, json={
    "session_id": session_id,
    "message_id": message_id,
    "high_quality": False
})
# Poll the new job_id...
```