AI Endpoint Generation
AI endpoint generation lets you describe what you want an endpoint to do in plain English, and Wrapd generates the full configuration for you. You review and edit the result before creating the endpoint.
How to use it
Section titled “How to use it”- Go to Dashboard → Endpoints and click Generate with AI
- Type a description of what you want — for example, “backup my postgres database every night at 2am”
- Click Generate and watch the AI stream its response
- Review the config preview that appears in the modal
- Refine — type a follow-up like “make it public” or “add a verbose flag” and click Refine to update the config
- Repeat until you’re happy, then click Use this config to open the create form pre-filled
- Make any final edits and click Create
You can go back and forth as many times as you need — the AI remembers the full conversation.
What AI generates
Section titled “What AI generates”The AI fills in all endpoint fields based on your description:
- Slug — a URL-friendly name
- Command — the shell command to run
- Method — GET, POST, etc.
- Description — what the endpoint does
- Parameters — named arguments with defaults
- Environment variables — any env vars the command needs
- Schedule — cron expression if you described a recurring task
- Timeout — based on expected runtime
- Access — public or private
Tips for good results
Section titled “Tips for good results”- Be specific about what tool or command you want to run
- Mention the schedule if you want it automated (“every hour”, “daily at midnight”)
- Include details about parameters callers should be able to pass
- Reference file paths or database names if relevant
Examples
Section titled “Examples”| Prompt | What you get |
|---|---|
| ”disk usage report for /data” | du -sh /data/* endpoint, JSON transformer |
| ”restart nginx” | systemctl restart nginx endpoint, POST method |
| ”backup postgres db every night at 2am” | pg_dump command with daily cron schedule |
| ”run pytest with optional verbose flag” | pytest with a verbose boolean parameter |
| ”check if a website is up” | curl -sI with a required url parameter |
Credits
Section titled “Credits”Each generation or refinement uses one AI credit. Credits reset monthly.
| Plan | AI generations / month |
|---|---|
| Free | 5 |
| Pro | 100 |
| Team | 500 |
Your remaining credits are shown in the top-right corner of the AI modal. When credits run out, the Generate/Refine button is disabled until the next billing cycle.
Safety
Section titled “Safety”Generated configs are validated before they reach you:
- Dangerous commands are blocked — destructive operations like recursive deletes, disk formatting, and piping untrusted URLs to a shell are rejected automatically
- Schema enforcement — only valid endpoint fields are accepted; unexpected keys are stripped
- Values are clamped — timeouts stay within 1–3600 seconds, methods are restricted to standard HTTP verbs
If the AI produces something unsafe, you’ll see an error instead of a config preview.
Always review the output
Section titled “Always review the output”AI generation is a starting point, not a finished product. Always check:
- The command is correct and exists on your machine
- File paths and database names match your setup
- The schedule expression is what you intended
- Access level (public/private) is appropriate
- Timeout is reasonable for the command’s expected runtime