Video Captioning by ReelWords

Generate captions for short-form videos using the ReelWords (reelwords.ai) Caption API. Use when a user asks to create caption jobs, render stylized subtitles for Reels/TikTok/Shorts, poll caption-job status, or download the rendered captioned video via the ReelWords REST API endpoints (/api/v1/caption-jobs) using an x-api-key (rw_...).

ReelWords Captions

Generate stylized captions for videos using the ReelWords Caption API: create a caption render job, poll until complete, then download the rendered output.

Setup

1) Create a ReelWords account + API key

1. Sign up / log in: https://reelwords.ai
2. Open the account menu (top-right) → API Keys → New key
3. Copy the key (it will look like rw_...).

2) Provide the API key to OpenClaw

Provide the API key to the process as REELWORDS_API_KEY.

Common options:

• Set REELWORDS_API_KEY as an environment variable (best when running the script directly)
• If you run via OpenClaw/Clawdbot, store it in ~/.clawdbot/openclaw.json under skills.entries.reelwords-captions.env so the runtime can populate env vars for the skill.

Option A: environment variable

bash

export REELWORDS_API_KEY="rw_..."

Option B: openclaw.json (recommended)

Edit ~/.clawdbot/openclaw.json and add an entry:

json

{
  "skills": {
    "entries": {
      "reelwords-captions": {
        "enabled": true,
        "env": {
          "REELWORDS_API_KEY": "rw_..."
        }
      }
    }
  }
}

Security note

Treat your API key like a password:

• don’t commit it to git
• don’t paste it into public chats
• rotate it in ReelWords if you suspect it leaked

Usage

Base URL: https://api.reelwords.ai

You can use either the included helper script (simplest), or call the REST endpoints directly.

Option 1: All-in-one helper script (create → poll → download)

From this skill directory:

bash

python3 scripts/reelwords_caption_job.py \
  --video-url "https://cdn.reelwords.ai/sample.mp4" \
  --style-id "style1" \
  --add-emojis \
  --max-words-per-line 6 \
  --position-y 82 \
  --font-size 54 \
  --highlight-color "#FFD803" \
  --hook-color "#FF5CAA" \
  --out captioned.mp4

Notes:

• The script prints the final job JSON to stdout.
• Download logic:
  • prefers result.downloadUrl when present
  • otherwise falls back to GET /api/v1/caption-jobs/{id}/video (which typically redirects to a signed URL)

Option 2: Raw API examples (curl)

1) Create job

bash

curl -sS https://api.reelwords.ai/api/v1/caption-jobs \
  -H "x-api-key: $REELWORDS_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "videoUrl": "https://cdn.reelwords.ai/sample.mp4",
    "preferences": {
      "style": {
        "styleId": "style1"
      }
    }
  }'

Response includes an id (save it as $JOB_ID).

2) Poll status

bash

curl -sS https://api.reelwords.ai/api/v1/caption-jobs/$JOB_ID \
  -H "x-api-key: $REELWORDS_API_KEY" \
  -H "accept: application/json"

Poll until status becomes completed (or the response includes failureReason/failureMessage).

3) Download rendered video

Preferred (when present):

• download from result.downloadUrl

Fallback (works in most tenants):

bash

curl -L https://api.reelwords.ai/api/v1/caption-jobs/$JOB_ID/video \
  -H "x-api-key: $REELWORDS_API_KEY" \
  -o captioned.mp4

Workflow (high level)

1. Create caption job: POST /api/v1/caption-jobs
   • videoUrl (required)
   • preferences.style.styleId (required)
   • optional preferences (emojis, max words per line, colors, font sizing, etc.)
2. Poll job status: GET /api/v1/caption-jobs/{id} until:
   • completed: result.downloadUrl is usually present (downloadable)
   • failed: failureReason / failureMessage present
3. Download:
   • result.downloadUrl, or
   • GET /api/v1/caption-jobs/{id}/video (redirects to a signed URL)

References

• API summary + curl examples: references/api.md

Notes

• Auth header is x-api-key: <token>.
• If you hit usage limits, treat HTTP 402 as “out of credits / limit reached” and surface the response cleanly.
• If ReelWords adds new fields, prefer passing them through in the JSON payload rather than hardcoding assumptions.