- Home
- Skills
- DevOps & Cloud
- Sentry Error Monitoring Setup and Audit for Any Stack
Works with the AI Agents you already use
Sentry Error Monitoring Setup and Audit for Any Stack
Set up or audit Sentry: official SDK, releases, CI source maps, PII and replay privacy, noise filters, throttled alerts, ownership, test error. Verdict MONITORED / NOISY / BLIND.
$5
Sentry Error Monitoring Setup and Audit for Any Stack
Example session with this skill installed
Audit our Sentry setup. We have a Vite React frontend and an Express API, both on @sentry v11, deployed with GitHub Actions. Stack traces in Sentry are minified and the team muted the alerts channel.
- Read your context and instructions
- Compiled the sentry error monitoring
Sentry setup: shop-web + shop-api — NOISY
Defaults used: SaaS, user ID only, errors-only replay, traces 0.1 in prod, alerts to #shop-alerts, GitHub Actions.
SDK: @sentry/react and @sentry/node 11.x (verified against docs.sentry.io today; v11 privacy defaults apply).
Findings
| # | Area | Status | Evidence | Fix |
|---|---|---|---|---|
| 1 | Source maps | Missing | vite.config.ts has no Sentry plugin | Add @sentry/vite-plugin, sourcemap: "hidden" |
| 2 | Release tag | Missing | src/sentry.ts:4 no release | Use git SHA from CI |
| 3 | PII | Over-collecting | dataCollection unset on v11 | Set userInfo: false, cookies: false |
| 4 | User context | Email sent | src/auth.ts:31 setUser({ email }) | Send { id } only |
| 5 | Noise | High | No denyUrls for extensions | Add extension filters |
| 6 | Alerts | Spam | Throttling = every trigger | 60 min, production only |
| 7 | API server | OK | instrument.mjs loaded with --import | none |
Changes
// vite.config.ts
build: { sourcemap: "hidden" },
plugins: [react(), sentryVitePlugin({
org: "acme", project: "shop-web",
authToken: process.env.SENTRY_AUTH_TOKEN,
sourcemaps: { filesToDeleteAfterUpload: ["./**/*.map"] },
})],
// src/sentry.ts
release: import.meta.env.VITE_SENTRY_RELEASE,
environment: import.meta.env.MODE,
dataCollection: { userInfo: false, cookies: false, httpBodies: [] },
denyUrls: [/^chrome(-extension)?:\/\//i, /^moz-extension:\/\//i],
CI (GitHub Actions)
env:
SENTRY_AUTH_TOKEN: ${{ secrets.SENTRY_AUTH_TOKEN }}
VITE_SENTRY_RELEASE: ${{ github.sha }}
- run: npx sentry-cli releases new "$VITE_SENTRY_RELEASE"
- run: npx sentry-cli releases set-commits --auto "$VITE_SENTRY_RELEASE"
Alerts (proposed, not created)
| Alert | Trigger | Throttling |
|---|---|---|
| New issue | A new issue is created, priority >= high | 60 min |
| Regression | A resolved issue regresses | 60 min |
| Spike | An issue escalates | 30 min |
Ownership: path:src/checkout/** #payments, path:api/** #backend.
Verification
Not yet run. After deploy to staging, open the temporary /debug-sentry route and confirm: original file:line in trace, release = commit SHA, no email in event JSON.
Open items
- Rotate nothing: no committed token found.
- Re-run audit after fixes; expected verdict MONITORED if the test error checks pass.
Connects securely to your tools. The creator never sees your data.
About this skill
Most Sentry installs work on day one and quietly decay. Stack traces stay minified because source maps never uploaded, every event says release unknown, extension errors drown real bugs, alerts fire on every event, and user emails and IPs end up in event payloads. The JavaScript SDK v11 also changed privacy defaults, so an unset option now collects more than it used to.
This skill has your coding agent set up or audit Sentry the way the current docs describe:
- Detects the stack (Next.js, React, Vue, Node/Express, Python/Django/FastAPI, Go, React Native, Flutter) and the installed SDK version
- Uses the official wizard or SDK install path and reviews what the wizard changed
- Sets release and environment consistently between the SDK and the upload step
- Uploads source maps or debug symbols from CI with bundler plugins or sentry-cli, and keeps maps off your public server
- Sets sampling (keep all errors, sample traces, drop health checks), errors-only Session Replay with masking on, and explicit PII settings (dataCollection or send_default_pii)
- Adds noise filters for browser extensions and known benign errors
- Proposes new issue, regression and spike alerts with throttling, ownership rules, CODEOWNERS and suspect commits
- Keeps the auth token in CI secrets and checks the repo for committed tokens
- Proves the pipeline with a test error and a checklist
Who it is for: developers and small teams adding Sentry, fixing unreadable traces, or cutting alert noise before a launch.
Honesty note: the skill tells the agent to verify SDK options and versions on docs.sentry.io before writing config, to label anything unverified, and not to claim the setup works until a test event is seen. It proposes alert and project changes but does not apply them without your approval.
The skill works from your code and CI config; it cannot read your Sentry organization settings unless you provide them. Alert and ownership changes are proposed, not applied. SDK defaults change between versions, so results depend on verifying the docs for your platform. Plan-gated features such as code owners import need a Business or Enterprise plan.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 29 days ago
Requires