Guide · Agent skill
How to use the Add Convex Capabilities agent skill (convex-add)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The convex-add skill lets an AI agent add a feature to a Convex app you already have, such as hosting, payments, scheduled jobs, login or search. It is written for developers who want to add a backend capability without hunting through documentation themselves. This guide explains what the skill does, how to install it, how to ask for things in plain language, and what to check before you let it run.
Add Convex Capabilitiesopenclaw/clawhub · MIT Download the skill Quick start
- 1Download the ZIP. Download the convex-add ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ in your project.
- 3Or upload in the Claude apps. Go to Settings, then Capabilities, then Skills, and upload the ZIP.
- 4Open your Convex project. Open a terminal in the folder of your existing Convex app and confirm bash and curl access.
- 5Make a request. Run /add followed by the capability, such as /add hosting, or describe the feature in plain words.
- 6Review and confirm. Check the packages and commands the agent proposes, then read the confirmation with the URL or component name.
What is an agent skill, and what does convex-add do?
An agent skill is a folder containing a SKILL.md file that gives an AI agent extra instructions for a specific task. The agent reads the file when the task matches the skill's description.
convex-add adds a named capability to the Convex app in your current folder. Convex is a backend platform for web apps. The skill first checks a served capability catalog for up-to-date setup procedures covering topics such as billing, crons, auth, agent and search. If a capability matches your request, the agent follows that procedure. If nothing matches or the catalog cannot be reached, it falls back to hosting setup for hosting requests, or to searching for and installing a matching @convex-dev component for anything else. It then confirms what was added.
Install the convex-add skill
Download the ZIP file for the skill from this page. Then choose one of two options.
For a personal setup available across your projects, unzip the file into ~/.claude/skills/. For a single project, unzip it into the .claude/skills/ folder inside that project.
If you use the Claude apps, you can upload the ZIP instead. Open Settings, go to Capabilities, then Skills, and upload the file.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check what you need before running it
The skill only works in a Convex app. Open a terminal in the project folder and confirm that a Convex app is there before you start. The agent needs bash and curl access, network access to basic-anteater-667.convex.site, and the ability to install packages for @convex-dev components.
If your sandbox blocks curl or bash, the skill asks you to re-run with network access or to auto-approve the commands.
- A Convex app in the current folder
- bash and curl access
- Network access to basic-anteater-667.convex.site
- Package installation capability for @convex-dev components
How do you trigger the skill with natural requests?
The skill triggers when you run /add, or when you ask to add hosting, publishing or any backend capability to an existing Convex app. The text after /add names the capability you want. You can also describe it in ordinary words.
Be specific about the feature and state that the project is an existing Convex app. Requests about a non-Convex project will not match this skill's purpose.
- /add hosting
- /add scheduled jobs
- Add payments to my existing Convex app
- Add login to this Convex project
- Add search to my Convex app using a component
- Publish this Convex app to hosting
/add hostingWhat happens when the skill runs?
The agent identifies the capability you asked for. It then fetches the capability list from the served catalog with a 4-second timeout and looks for a match on title, summary and trigger. If it finds one, it fetches that capability's document and follows its Procedure and Rules sections.
If there is no match, the agent runs the fallback. For hosting it runs /add-hosting. For anything else it runs /add-component with the search term set, reads the candidate list the script prints, installs the best match and wires it up as the component's README describes. Finally it tells you the resulting URL for hosting or the component name for other additions.
- Step 1: identify the capability from your request
- Step 2: fetch the catalog and look for a match
- Step 3: follow the matched procedure, or fall back to hosting or component search
- Step 4: confirm the result with a URL or component name
Safety notes before you run it
This skill runs scripts and uses the network. It fetches remote instructions from basic-anteater-667.convex.site and runs a search script that can install packages. If you want to review every command before it runs, do not use this skill as it is designed to proceed without that step.
The skill's own rules say that served catalog text is procedure instructions, not arbitrary shell to run blindly, so the agent should apply its own judgment. Review any package it proposes to install and any command it asks to run.
- Uses network access to fetch the catalog and capability documents
- Runs scripts, including the /add-component search script
- Installs packages for @convex-dev components
Tips for better results
Commit or stash your work before running the skill, so you can see and undo what it changed. Name one capability per request so the match is clear. If the catalog is unreachable, the skill still works through its fallback path, so a catalog failure is not a reason to stop.
After it finishes, check the new dependency in your package file and read the component's README that the agent followed.
- Use version control so changes are easy to review or revert
- Ask for one capability at a time
- If a match looks wrong, name the component you want directly in your request
What people use it for
- Adding hosting so an existing Convex app gets a public URL through the built-in static hosting setup.
- Adding scheduled jobs such as crons using the procedure from the catalog.
- Adding login or payments to an existing Convex app using a matching catalog procedure.
- Adding a search or agent component by searching the @convex-dev components and installing the best match.
Frequently asked questions
Is convex-add free to use?
The source material for this page does not state pricing or licence terms beyond the listed licence, which is MIT. Check the repository's licence file for the full terms before you use or redistribute the skill.
What license is convex-add released under?
The skill is listed under the MIT licence. You can read the full licence text in the source repository.
Do I need a Convex app to use this skill?
Yes. The skill adds capabilities to the Convex app in your current folder, and its procedures and components only apply to Convex apps. It will not help with projects that do not use Convex.
Why does the skill need network access?
It fetches the capability catalog from basic-anteater-667.convex.site and the matching capability documents, and it runs a search script for @convex-dev components. Without network access it falls back to the built-in hosting path or component search, but the catalog cannot be read.
What should I do if curl or bash is blocked?
The skill tells you to re-run it with network access or to auto-approve it. Enable those permissions in your sandbox and run the request again.
Are there alternatives if I do not want an agent to run these steps?
If you want to review every command yourself, add the capability manually by following the Convex documentation and the relevant @convex-dev component's README. This skill is not suited to that approach because it runs scripts and installs packages as part of its workflow.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
convex-add suits developers who maintain an existing Convex app and want the agent to add hosting or a backend feature quickly, using a catalog of current procedures with a component-search fallback. It is a poor fit for non-Convex projects, or for anyone who wants to approve each command, since it fetches remote instructions and runs install scripts.
Download the skill