Skip to content
Get packs

Convex Deployment Guard

Makes the agent identify and announce the Convex target before deploy-affecting commands, and gate prod changes behind fresh consent.

Source Download skill
Repository
openclaw/clawhub
Stars
9,546
Licence
MIT
Files
1 file

What it does

Convex apps often have several deployments (personal dev, preview, production), and a command sent to the wrong one is a common source of mistakes. With this skill installed, the agent first checks which deployment it is pointing at, states that target in one line, and only then runs deploy, env, import/export or similar commands. Any change to production requires a fresh, explicit yes for that specific action, and a session can be locked to read-only.

Great for

  • Developers who work on several Convex projects or deployments on one machine
  • Teams that want guardrails so an AI agent cannot accidentally deploy or mutate production
  • People who want a read-only audit of production data without enabling production changes

Skip it if

  • Projects that have no Convex backend, since the workflow is specific to Convex deployments
  • Situations where you want the agent to deploy to production without confirmations; the skill intentionally blocks this by design
  • Anyone expecting it to perform deployments itself: per its own rules it is a safety discipline that other deploy skills call as a first step, not a deploy tool

Try asking

  • “Deploy my changes to Convex, but first tell me exactly which deployment you are targeting”
  • “Put this session in read-only mode, don't change anything in the database”
  • “Audit production logs for errors without changing anything in prod”
  • “My deploy said success but the app didn't change, figure out which deployment it hit”
Step-by-step guide: how to use Convex Deployment GuardInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    convex-deploy-guard.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip convex-deploy-guard.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip convex-deploy-guard.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Ask for read-only mode at the start of a session if you only want investigation; the agent treats it as absolute for that session
  • Give explicit, specific approval for production changes, naming the deployment and the change; earlier approvals do not carry over
  • Use this skill alongside other Convex deploy or migration skills, since they are designed to run it as their first step

What's inside

  • SKILL.md

Redistributed unmodified from openclaw/clawhub under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.