openclip-captions
Burn styled, animated captions into a short clip with OpenClip using caption presets (beast, pop, kendrick, mozi, dan, sara, lucy, tayo and more) or a custom caption style and position. Use when asked to "add captions", "add subtitles to this short", "burn in the text", "make it look like MrBeast captions", "restyle the captions", "put words on screen", or "caption this clip".
How do I install this agent skill?
npx skills add https://github.com/openclip-app/agent-skills --skill openclip-captionsIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill is safe and provides instructions for an AI agent to use the OpenClip service to add animated captions to video clips. It uses the Model Context Protocol (MCP) to connect to the vendor's official API.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
OpenClip: captions on clips
render_clip burns animated captions into a viral moment produced by the OpenClip clipping
pipeline. This is part of the PAID pipeline, it needs an active subscription and credits.
Captions here are rendered into the video, not a sidecar subtitle file. If the user wants an
SRT or VTT file instead, that is the free transcribe tool, see openclip-transcription.
Setup (once)
OpenClip is a remote MCP server at https://openclip.app/mcp. Sign in with your OpenClip
account when the OAuth prompt appears, there is no API key to copy.
- Claude Code:
claude mcp add --transport http openclip https://openclip.app/mcp, then/mcpto authorize. - Claude Desktop / web: Settings, Connectors, "Add custom connector", paste the URL, sign in.
- Cursor (
.cursor/mcp.json):{ "mcpServers": { "openclip": { "url": "https://openclip.app/mcp" } } } - Header-only clients: mint an MCP token at
openclip.app/settings/connectand connect tohttps://openclip.app/mcp/keywithAuthorization: Bearer <token>.
Sanity check: call get_account and confirm credits_remaining > 0.
The loop
- Have a moment. You need a
moment_idfromlist_clips(job_id). If the user has not run a video through the pipeline yet, start withsubmit_video, see theopenclip-clippingskill. - Pick a preset. Call
list_caption_presetsto get the current keys. Known keys:default,dan,dan-reveal,pop,mozi,kendrick,sara,lucy,tayo,beast. Always confirm with the call, keys can change. That same call also returns the full caption-style property schema and the caption position options, which is what you need for custom styling rather than a named preset. - Render.
render_clip(moment_id, caption_preset). - Poll. Rendering is async. Poll
list_clips(job_id)and watch that moment'srendered_clipfield. When it appears it holds{url, size_bytes}, the finished captioned video.get_render_statusreports on the render directly.
Choosing a preset
Presets are visual styles, not just fonts. When the user names a creator or a vibe, map it:
- Big, punchy, word-by-word pop with heavy outline -
beast. - Clean and readable, safe default for a brand account -
default. - Bouncy and playful -
pop. - Reveal-style where words appear progressively -
dan-reveal.
When the user does not care, use default and show them the other keys so they can swap.
When they want something no preset covers, pull the caption-style property schema from
list_caption_presets and build a custom style instead of forcing a near-miss preset.
Rules
- Always confirm preset keys with
list_caption_presetsbefore rendering. An unknowncaption_presetcomes back as a validation message on the tool result. - Always poll.
render_clipreturns immediately, the video is not ready yet. - A moment can be re-rendered with a different preset.
rendered_clipreflects the latest successful render. clipis the unrendered cut,clip_watermarkedis the watermarked version,rendered_clipis the captioned output. Hand the user the right one.- Times are milliseconds.
Errors
Tool failures come back as a tool result with isError: true and a plain-text message, not an
HTTP status.
"Viral moment not found."- themoment_idis wrong or belongs to another account. Re-fetch withlist_clipsand use an id from that response."Not authenticated. Reconnect your OpenClip token."- reconnect or re-mint the token.- An unknown preset key returns the validation message text. Call
list_caption_presetsand retry with a real key.
Example prompts to actions
- "Add captions to that clip" -
list_caption_presets,render_clip(moment_id, caption_preset="default"), poll. - "Make it look like MrBeast" -
render_clip(moment_id, caption_preset="beast"), poll. - "What caption styles are there?" -
list_caption_presets, show the keys. - "Caption all the top 3" -
list_clips, take the top 3 byvirality_score,render_clipeach, poll until everyrendered_clipis present. - "I want the captions at the top in my brand font" -
list_caption_presetsfor the style schema and position options, then render with a custom style.
Related
Producing the clips to caption is openclip-clipping. Standalone SRT and VTT subtitle files are
openclip-transcription. Locking a caption preset into a reusable processing agent is covered
in openclip-clipping.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/openclip-app/agent-skills/openclip-captions">View openclip-captions on skillZs</a>