convex-sentinel Skill for Claude Code: What It Does and How to Install It

Makes Claude set up Sentinel in your Convex project so production errors are saved in your own deployment.

Installs: 38.2K; on skills.sh
Stars: 68; on GitHub
Updated: Oct 2026; last repo commit

This skill makes Claude install @convex-dev/sentinel, a Convex component that captures production errors into a table in your own deployment. It covers server function failures, client JavaScript and React crashes, and conflict and scale signals. Secrets are redacted when the error is written, and that is on by default.

Claude adds app.use(sentinel) to convex/convex.config.ts and wires the client with a React error boundary, window.onerror and unhandledrejection handlers, and breadcrumbs. You then read errors with the Convex CLI or let convex-monitor react to them. Capturing prod errors needs a deployed cloud app, while the install works anonymously. Its folder holds only SKILL.md.

Good for

Convex developers who want production errors captured in their own backend
Teams that need secrets redacted before errors are stored

Install convex-sentinel

Claude Code

Copy this prompt.

Install the convex-sentinel skill in Claude Code. Run this command:

npx skills add https://github.com/convex/agent-skills --skill convex-sentinel

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

SKILL.md names no other skill it needs. It also mentions convex-monitor and convex-self-heal as optional hand-offs; install those separately if you want them. Convex's docs say the skills CLI installs skills into .agents/skills/ in your project. The "All Convex skills" command installs every skill in the repo, which has 33 skill folders. One of them, convex-improve-convex-plugin, is listed in the Convex docs as sending your coding session's transcript to the Convex team. Run this skill with /convex-sentinel in Claude Code or $convex-sentinel in Codex, or let the agent pick it when the task fits. This skill works on your project folder, so this page offers no Claude app upload route.

Built from the package data we hold. Not yet tested by us.

What's in the skill

When it loads

Set up Sentinel production error capture in your own Convex deployment.

Sections in SKILL.md

Workflow
Rules

Key facts

Author: Convex (GitHub organisation convex; the vendor's own repo)
License: Apache-2.0 (stated in the README and the LICENSE file)
Installs on skills.sh: 38,218 (Nov 2, 2026)
GitHub stars: 68 for the whole convex/agent-skills repo (Nov 2, 2026)
Last repo commit: Oct 29, 2026
Audits on skills.sh: Gen Agent Trust Hub: pass, Socket: pass, Snyk: warn
What it includes: 1 file: SKILL.md (the repo tree lists nothing else in the folder)
Part of: The convex/agent-skills repo: 33 skill folders in the tree. The README names 6 skills, and this one is not among them

What it includes

Adds app.use(sentinel) to convex/convex.config.ts
The component package is @convex-dev/sentinel.
Runs code · Install the component
Adds a React error boundary, window.onerror and unhandledrejection handlers
Write · Wire the client
On by default; denies secret key names and value patterns before storing
SKILL.md says never to store raw secrets.
Redaction at write time
Uses convex data or a run-once query through the Convex CLI
Runs code · Read recent errors
Hands new ones to convex-monitor through its prod_error event
React to new errors
A triage step classifies each error and hands recurring ones to ai-runner to open a fix PR
SKILL.md does not define ai-runner. The full fix loop is convex-self-heal.
Write · Optional self-healing cron

From the official docs

Permissions

Medium risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Errors are stored in your database
Captured errors sit in a table in your deployment. SKILL.md says redaction is mandatory and on by default, and that the agent's reads reach the model provider. Check what your error payloads contain. github.com
The optional cron can open fix PRs
The self-healing cron hands recurring non-transient errors to ai-runner to open a fix PR. SKILL.md marks it optional, so enable it deliberately. github.com
It adds code to your client app
Error boundaries and global handlers change how your app reports crashes. Review the diff. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	SKILL.md defines no read-only mode.
Sign-in and keys	SKILL.md names no credentials. It says redaction denies secret key names and value patterns by default.
Scopes	Edits convex/convex.config.ts and your client code Stores captured production errors in a table in your own Convex deployment Reads recent errors through the Convex CLI Optionally enables a cron that can lead to fix PRs

Questions

What does convex-sentinel do?

It makes Claude set up Sentinel in your Convex app. Sentinel saves production errors from your server functions and your client code into a table in your own deployment, so Claude can read and react to them later.

Where do the errors go?

SKILL.md says the data stays in your deployment and must never be sent to a third party. It also says the agent's reads reach the model provider, which is why redaction is mandatory and on by default.

What gets redacted?

SKILL.md describes default-deny redaction at write time on secret key names and value patterns. It does not list every pattern, so check what your errors contain.

Do I need a deployed app?

To capture production errors, yes. SKILL.md says capturing prod errors needs a deployed cloud app, while the install itself works anonymously.

How do I install it?

Use the install box at the top of this page. Add convex-monitor separately if you want Claude to react to new errors, and convex-self-heal for the full fix loop. Run it with /convex-sentinel in Claude Code or $convex-sentinel in Codex.

Is it safe?

It edits your Convex config and client code and starts storing error data in your deployment. SKILL.md says to sample and cap volume to control cost. The optional cron can lead to fix PRs being opened, so enable it only when you want that.

What is the optional self-healing cron?

A triage step that classifies each error and hands recurring, non-transient ones to ai-runner to open a fix PR. SKILL.md does not describe ai-runner further. convex-self-heal is the separate skill for the full certified-fix loop.

Is it official?

It is in the convex/agent-skills repo under the Convex GitHub organisation, which is the vendor's own repo.

Information

updated Nov 2, 2026

Repository: convex/agent-skills
Folder: skills/convex-sentinel

convex/agent-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

convex/agent-skills README · github.com
skills CLI commands · github.com
Convex docs: Agent Skills · docs.convex.dev
convex-sentinel SKILL.md · github.com