Skip to main content
Each recipe shows the full API flow for a specific job. Pick the one closest to what you’re building and adapt it.

​
Podcast production pipeline

The job: You record a podcast episode and want it cleaned up and captioned without opening Descript.
1

Import the recording

curl -X POST https://descriptapi.com/v1/jobs/import/project_media \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "project_name": "Episode 42 - API Launch",
    "add_media": {
      "recording.mp4": {
        "url": "https://your-storage.com/episode-42.mp4"
      }
    },
    "add_compositions": [
      { "name": "Full Episode", "clips": [{ "media": "recording.mp4" }] }
    ]
  }'
2

Wait for import to complete

Poll GET /jobs/{job_id} until job_state is "stopped", or use a callback_url.
3

Clean up and caption

curl -X POST https://descriptapi.com/v1/jobs/agent \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "project_id": "YOUR_PROJECT_ID",
    "prompt": "Remove filler words, apply Studio Sound, detect and label speakers, and add captions"
  }'
4

Review and export

Open the project in Descript to review the edits and export the final episode.
Total API calls: 3 (import, poll, agent edit) Credits used: Media minutes for import + AI credits for agent edit

​
Social clip factory

The job: You have a long-form recording (webinar, interview, lecture) and want short highlight clips for social media.
1

Import the long-form video

curl -X POST https://descriptapi.com/v1/jobs/import/project_media \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "project_name": "Webinar Highlights - April",
    "add_media": {
      "webinar.mp4": {
        "url": "https://your-storage.com/webinar-full.mp4"
      }
    },
    "add_compositions": [
      { "name": "Full Recording", "clips": [{ "media": "webinar.mp4" }] }
    ]
  }'
2

Wait for import, then create highlights

curl -X POST https://descriptapi.com/v1/jobs/agent \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "project_id": "YOUR_PROJECT_ID",
    "prompt": "Create three 30-second highlight clips from the most interesting moments. Add captions to each clip."
  }'
3

Review and export clips

Open in Descript. The agent creates new compositions for each clip. Export them individually.

​
Training video batch processing

The job: Your team records training videos regularly. Each one needs the same cleanup — audio enhancement, filler removal, captions. 
import requests
import time

API_TOKEN = "YOUR_API_TOKEN"
BASE = "https://descriptapi.com/v1"
HEADERS = {"Authorization": f"Bearer {API_TOKEN}", "Content-Type": "application/json"}

# List of training recordings to process
recordings = [
    ("Onboarding Module 1", "https://storage.com/onboarding-1.mp4"),
    ("Onboarding Module 2", "https://storage.com/onboarding-2.mp4"),
    ("Onboarding Module 3", "https://storage.com/onboarding-3.mp4"),
]

STANDARD_EDIT = "Remove filler words, apply Studio Sound, and add captions"

for name, url in recordings:
    # Import
    resp = requests.post(f"{BASE}/jobs/import/project_media", headers=HEADERS, json={
        "project_name": name,
        "add_media": {"video.mp4": {"url": url}},
        "add_compositions": [{"name": "Main", "clips": [{"media": "video.mp4"}]}]
    })
    job = resp.json()
    print(f"Importing: {name}")

    # Wait for import
    while True:
        status = requests.get(f"{BASE}/jobs/{job['job_id']}", headers=HEADERS).json()
        if status["job_state"] == "stopped":
            break
        time.sleep(10)

    # Apply standard edits
    resp = requests.post(f"{BASE}/jobs/agent", headers=HEADERS, json={
        "project_id": job["project_id"],
        "prompt": STANDARD_EDIT
    })
    print(f"Editing: {name} → {resp.json()['job_id']}")
This processes each recording with identical edits. Change STANDARD_EDIT to match your team’s standard processing.

​
Content repurposing

The job: Turn one recording into multiple outputs — a clean full version, a highlights reel, and a captioned version for social.
1

Import the source recording

Import as usual with POST /jobs/import/project_media.
2

Apply base cleanup

{
  "project_id": "YOUR_PROJECT_ID",
  "prompt": "Remove filler words and apply Studio Sound"
}
This gives you a clean base to work from.
3

Create a highlight reel

{
  "project_id": "YOUR_PROJECT_ID",
  "prompt": "Create a 60-second highlight reel from the best moments and add captions"
}
4

Review all outputs in Descript

The project now has your original composition (cleaned up) plus a new highlight composition. Open in Descript to review and export each version.

​
What all recipes have in common

Every workflow follows the same pattern:
  1. Import → POST /jobs/import/project_media
  2. Wait → Poll GET /jobs/{job_id} or use callback_url
  3. Edit → POST /jobs/agent with a specific prompt
  4. Wait → Poll again or use callback
  5. Review → Open project_url in Descript to export
The difference is what you put in the prompt. See the Prompt Guide for how to write prompts that get consistent results.