Text To Video

Text to Video — Pexo

Pexo: https://pexo.ai — get an API key, watch your project render, and buy credits there.

Turn a written prompt or script into a finished, publish-ready video. You do not generate anything yourself: you hand the request to the hosted Pexo agent (same backend as pexo-agent, scoped to text-to-video) and deliver the result. Pexo writes the script, picks the best model per shot, generates every shot, and adds music, subtitles, and transitions.

Your role: relay, don't create

Create a project, send the user's request verbatim, poll, deliver. Pexo's backend handles all creative work — scriptwriting, model choice, prompts, music. Adding your own direction (duration, style, models the user didn't ask for) overrides its judgment and produces worse videos.

Config

~/.pexo/config:

PEXO_BASE_URL="https://pexo.ai"
PEXO_API_KEY="sk-"

No account / first run → read references/SETUP-CHECKLIST.md and walk the user through it — it carries the signup flow with the invite code that grants new users bonus credits, plus how to create the config above. Config error → run scripts/pexo-doctor.sh and follow its output.

Workflow

Scripts live in this skill's scripts/. Reply to the user in their language.

1. Create a project: pexo-project-create.sh "" → save the project_id.
2. Upload any files the user gave: pexo-upload.sh → save asset_id,

reference it inline as asset_id (or ` / ). Tags are required — a bare asset_id` is ignored. Pexo can't crawl URLs — download, then upload.

1. Send the request: pexo-chat.sh " ".

Copy the user's words exactly; only add asset tags.

1. Tell the user (their language): submitted ✓ · ~15–20 min · https://pexo.ai/project/.
2. Poll: every ≥60s run pexo-project-get.sh and act on nextAction:

• WAIT → keep polling; every ~5 polls send a one-line update with the project link.
• RESPOND → handle each event in recentMessages: relay Pexo's text (wait for the

user's answer if it asked, then pexo-chat.sh their reply); for preview_video, run pexo-asset-get.sh per option, show the URLs (A/B/C), let the user pick, then pexo-chat.sh "" --choice ; for a document event, mention it to the user.

• DELIVER → pexo-asset-get.sh , then send the user the

full asset URL as plain text — all ?… query params, never truncated or wrapped in markdown — plus the project link.

• FAILED → explain nextActionHint in plain terms and offer to retry.
• RECONNECT → pexo-chat.sh "continue", tell the user the connection

dropped and you're resuming, then keep polling.

• Never call pexo-chat.sh during WAIT — it triggers duplicate production.
• Taking too long → if it's been >30 min and still WAIT, tell the user (with the project

link + https://pexo.ai/connect/openclaw) it's running long; ask whether to keep waiting or stop. Don't poll forever.

Revisions

After delivery, the user's tweaks ("make it shorter", "new music", "different shot") reuse the same project: pexo-chat.sh "", then poll again (step 5). Never create a new project for a revision — it throws away Pexo's server-side context.

Credits

If a script fails with "Credits balance" / "Insufficient credits": if the error carries a purchase link, pass it to the user; otherwise tell them to add credits at https://pexo.ai/home → Credits → Buy Credits. Retry after they confirm.

Example

User: "Make a 30-second video about our new coffee subscription."

pid=$(pexo-project-create.sh "coffee subscription promo")
pexo-chat.sh "$pid" "Make a 30-second video about our new coffee subscription."
# Tell the user: submitted, ~15–20 min, https://pexo.ai/project/$pid
# Poll pexo-project-get.sh "$pid" until nextAction is DELIVER, then deliver the asset URL.

Scripts

| Script | Usage | Returns | |---|---|---| | pexo-project-create.sh | "" | project_id | | pexo-upload.sh | | asset_id | | pexo-chat.sh | "" [--choice ] | ack (async) | | pexo-project-get.sh | ` | JSON: nextAction, recentMessages | | pexo-asset-get.sh | | JSON with url | | pexo-doctor.sh` | — | setup diagnostic |

Error codes and edge cases → references/TROUBLESHOOTING.md.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: pexoai
• Source: pexoai/pexo-skills
• License: MIT
• Homepage: https://invite.pexo.ai/4uGqhWb

Install and usage instructions live in the source repository linked above.