Capture production errors into a Convex deployment with write-time redaction.
npx skills add https://github.com/openclaw/clawhub --skill convex-sentinel
Please help me install this Agent Skill. Skill: convex-sentinel Source: https://github.com/openclaw/clawhub/tree/main/.agents/skills/convex-sentinel Command: npx skills add https://github.com/openclaw/clawhub --skill convex-sentinel
Production errors in Convex apps—server function failures, client JS/React crashes, OCC and scale signals—are hard to observe without sending data to third-party monitoring services. This Skill sets up Sentinel so errors are captured into a table in your own deployment, redacted at write time.
@convex-dev/sentinel via app.use(sentinel) in convex/convex.config.ts so error data never leaves your own deployment.window.onerror/unhandledrejection handlers and breadcrumbs to capture client-side crashes.convex data, run-once-query), react to new ones via the monitor's prod_error event, and optionally enable a self-healing cron that triages recurring errors and hands them to ai-runner to open a fix PR.Set up Sentinel error capture in my Convex deployment by installing the component, wiring the React client SDK, and showing me how to query recent production errors.
High-intent search queries and answers about installing and using this skill.
Install the @convex-dev/sentinel component with app.use(sentinel) in convex/convex.config.ts, then wire the client SDK with a React error boundary and window.onerror handlers. Errors are stored in a table in your own deployment and readable via the Convex CLI.
Wire the Sentinel client SDK by adding a React error boundary plus window.onerror and unhandledrejection listeners with breadcrumbs. Client crashes are captured into the same deployment table as server function failures.
No, all captured error data stays in your own Convex deployment and is never sent externally. Redaction runs at write time with default-deny rules on secret key names and value patterns.
Yes, an optional self-healing cron classifies each error via triage and, for recurring non-transient ones, hands it to ai-runner to open a fix PR. This is opt-in and disabled by default.
Capturing production errors requires a deployed cloud app, though installation works anonymously. You should also sample and cap event volume to control storage and cost.
wixelsAudit Convex applications to diagnose and resolve performance bottlenecks.
ReviewStageMonitor open pull requests and apply fixes for review comments, CI failures, and merge conflicts.
genesis-agentsDiagnose production issues by analyzing Railway logs and traces.
richardadonnellConfigure Sentry structured logging for JavaScript, Python, and Ruby projects.