Skip to main content ホーム クリエイター jeremylongshore tons-of-skills-marketplace sentry-reference-architecture
sentry-reference-architecture Design production-grade Sentry architecture for multi-service organizations.
Use when planning Sentry rollout, structuring projects across teams,
building shared config modules, or setting up distributed tracing.
Trigger: "sentry architecture", "sentry project structure",
"sentry reference design", "sentry distributed tracing".
インストールへ移動 Skills Marketplace コミュニティが作成したAIスキルを発見・探索
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace --skill sentry-reference-architectureコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Zipをダウンロード ダウンロード中... このリポジトリの他の Skills langchain-deploy-integration Deploy a LangChain 1.0 / LangGraph 1.0 app to Cloud Run, Vercel, or LangServe correctly — with timeouts sized for chain length, cold-start mitigation, SSE anti-buffering headers, and Secret Manager over .env. Use when prepping a first production deploy, debugging a stream that hangs behind a proxy, or diagnosing p99 latency spikes. Trigger with "langchain deploy", "langchain cloud run", "langchain vercel python", "langchain langserve", or "langchain docker".
langchain-langgraph-agents Build a correct LangGraph 1.0 ReAct agent with create_react_agent — typed tools, error propagation, recursion caps, and stop conditions that actually stop. Use when writing a first tool-calling agent, migrating from AgentExecutor or initialize_agent, or diagnosing an agent that loops on vague prompts. Trigger with "langgraph agent", "create_react_agent", "langgraph tool calling", "AgentExecutor migration", or "agent loop cost".
langchain-langgraph-human-in-loop Build LangGraph 1.0 human-in-the-loop approval flows with interrupt_before /
interrupt_after and Command(resume=...) — JSON-serializable state, clean
resume semantics, and UI wiring for approval decisions. Use when adding an
approval gate before an expensive tool call, wiring a Slack/web UI for agent
approvals, or debugging a graph that crashes on interrupt.
Trigger with "langgraph human in loop", "langgraph interrupt_before",
"langgraph approval flow", "Command resume", "langgraph HITL".
name sentry-reference-architecture description Design production-grade Sentry architecture for multi-service organizations.
Use when planning Sentry rollout, structuring projects across teams,
building shared config modules, or setting up distributed tracing.
Trigger: "sentry architecture", "sentry project structure",
"sentry reference design", "sentry distributed tracing".
allowed-tools Read, Write, Edit, Bash(npm:*), Bash(npx:*), Glob, Grep version 1.51.0 license MIT author Jeremy Longshore <jeremy@intentsolutions.io> tags ["saas","sentry","architecture","enterprise","distributed-tracing","microservices"] compatibility Designed for Claude Code
Sentry Reference Architecture
Overview
Enterprise Sentry architecture patterns for multi-service organizations. Covers centralized configuration, project topology, team-based alert routing, distributed tracing, error middleware, source map management, and a production-ready SentryService wrapper.
Prerequisites
Sentry organization at sentry.io (Business plan+ for team features)
@sentry/node v8+ installed (npm install @sentry/node @sentry/profiling-node)Service inventory and team ownership documented
Node.js 18+ (ESM and native fetch instrumentation)
Instructions
Step 1 — Project Structure Strategy
Pattern A: One Project Per Service (3+ services, recommended)
Organization: acme-corp
├── Team: platform-eng
│ ├── Project: api-gateway (Node/Express)
│ ├── Project: auth-service (Node/Fastify)
│ └── Project: user-service (Node/Express)
├── Team: payments
│ ├── Project: payment-api (Node/Express)
│ └── Project: billing-worker (Node worker)
└── Team: frontend
├── Project: web-app (React/Next.js)
└── Project: mobile-app (React Native)
Benefits: independent quotas, team-scoped alerts, per-service rate limits, isolated release tracking.
Pattern B: Shared Project (< 3 services, single team) — one project with Environment tags (production/staging/dev). Simpler setup; outgrow when alert noise exceeds one team.
Step 2 — Centralized Config Module
Create lib/sentry.ts imported by every service to enforce org-wide defaults:
import * as Sentry from '@sentry/node' ;
import { nodeProfilingIntegration } from '@sentry/profiling-node' ;
export interface SentryServiceConfig {
serviceName : string ;
dsn : string ;
environment ?: string ;
version
string
tracesSampleRate
number
ignoredTransactions
string
export
function
initSentry
config : SentryServiceConfig
void
const
environment
env
NODE_ENV
'development'
Sentry
init
dsn
dsn
environment
release
`${config.serviceName} @${config.version || 'unknown' } `
serverName
serviceName
tracesSampleRate
tracesSampleRate
'production'
0.1
1.0
sendDefaultPii
false
maxBreadcrumbs
50
integrations
nodeProfilingIntegration
ignoreErrors
'ResizeObserver loop completed with undelivered notifications'
/Loading chunk \d+ failed/
'AbortError'
tracesSampler
({ name, parentSampled } ) =>
if
undefined
return
const
ignoredTransactions
'GET /health'
'GET /healthz'
'GET /ready'
'GET /metrics'
if
some
p =>
includes
return
0
return
tracesSampleRate
'production'
0.1
1.0
beforeSend
event
if
request
headers
delete
request
headers
'authorization'
delete
request
headers
'cookie'
delete
request
headers
'x-api-key'
return
initialScope
tags
service
serviceName
team
env
TEAM_NAME
'unassigned'
export
Sentry
Bootstrap in each service:
import { initSentry } from '@acme/sentry-config' ;
initSentry ({ serviceName : 'api-gateway' , dsn : process.env .SENTRY_DSN ! });
Step 3 — Error Handling Middleware
import * as Sentry from '@sentry/node' ;
import type { Request , Response , NextFunction , ErrorRequestHandler } from 'express' ;
export function sentryRequestHandler (return (req : Request , _res : Response , next : NextFunction ): void =>Sentry .setTag ('http.route' , req.route ?.path || req.path );
if (req.user ) Sentry .setUser ({ id : req.user .id });
next ();
};
}
export const sentryErrorHandler : ErrorRequestHandler = (err, req, res, _next ) => {
const status = (err as any ).statusCode || 500 ;
Sentry .withScope ((scope ) => {
scope.setLevel (status >= 500 ? 'error' : 'warning' );
scope.setTag ('http.status_code' , String (status));
scope.setContext ('request' , { method : req.method , url : req.originalUrl });
status >= 500 ? Sentry .captureException (err) : Sentry .captureMessage (err.message , 'warning' );
});
res.status (status).json ({ error : status >= 500 ? 'Internal server error' : err.message });
};
import sentry_sdk, os
from sentry_sdk.integrations.fastapi import FastApiIntegration
def init_sentry (service_name: str ) -> None :
sentry_sdk.init(
dsn=os.environ["SENTRY_DSN" ],
environment=os.getenv("ENV" , "development" ),
release=f"{service_name} @{os.getenv('SERVICE_VERSION' , 'unknown' )} " ,
traces_sample_rate=0.1 ,
send_default_pii=False ,
integrations=[FastApiIntegration(transaction_style="endpoint" )],
before_send=lambda event, hint: _scrub(event),
)
def _scrub (event ):
headers = event.get("request" , {}).get("headers" , {})
for k in ["authorization" , "cookie" , "x-api-key" ]:
headers.pop(k, None )
return event
Step 4 — Distributed Tracing HTTP (automatic): SDK v8 auto-propagates sentry-trace + baggage headers on fetch/http. All services in the same org link automatically.
Message queues (manual propagation):
import * as Sentry from '@sentry/node' ;
export function publishWithTrace<T>(queue : string , payload : T, publish : Function ) {
return Sentry .startSpan ({ name : `queue.publish.${queue} ` , op : 'queue.publish' }, async () => {
const headers : Record <string , string > = {};
const span = Sentry .getActiveSpan ();
if (span) {
headers['sentry-trace' ] = Sentry .spanToTraceHeader (span);
headers['baggage' ] = Sentry .spanToBaggageHeader (span) || '' ;
}
await publish (queue, { payload, headers });
});
}
export function consumeWithTrace<T>(queue : string , msg : { payload : T; headers : Record <string , string > }, handler : (p : T Promise <void >) {
return Sentry .continueTrace (
{ sentryTrace : msg.headers ['sentry-trace' ], baggage : msg.headers ['baggage' ] },
() => Sentry .startSpan ({ name : `queue.process.${queue} ` , op : 'queue.process' }, () => handler (msg.payload ))
);
}
Step 5 — Team-Based Alert Routing Configure in Project Settings > Ownership Rules :
path:src/payments/**
path:src/auth/**
url:*/api/v1/payments/*
tags.service:payment-api
*
P0 Critical: error rate > 50/min OR crash-free < 95% — PagerDuty on-call, 15 min SLAP1 Warning: new production issue or regression — Slack #alerts-prod, same-day SLAP2 Performance: P95 > 2s or Apdex < 0.7 — Slack #alerts-perf, next sprintP3 Info: new staging issue — Slack #alerts-staging, backlog triage
Step 6 — Source Map Uploads (Monorepo) #!/usr/bin/env bash
set -euo pipefail
SERVICE="${1:?Usage: upload-sourcemaps.sh <service>} "
RELEASE="${SERVICE} @$(git rev-parse --short HEAD) "
npx @sentry/cli releases new "$RELEASE " --org "$SENTRY_ORG " --project "$SERVICE "
npx @sentry/cli sourcemaps upload --org "$SENTRY_ORG " --project "$SERVICE " \
--release "$RELEASE " --url-prefix "~/" --validate "./services/${SERVICE} /dist/"
npx @sentry/cli releases set-commits "$RELEASE " --org "$SENTRY_ORG " --auto
npx @sentry/cli releases finalize "$RELEASE " --org "$SENTRY_ORG "
Webpack plugin alternative (auto-uploads on build):
import { sentryWebpackPlugin } from '@sentry/webpack-plugin' ;
export default {
devtool : 'source-map' ,
plugins : [sentryWebpackPlugin ({
org : process.env .SENTRY_ORG ,
project : 'web-app' ,
authToken : process.env .SENTRY_AUTH_TOKEN ,
sourcemaps : { filesToDeleteAfterUpload : ['./dist/**/*.map' ] },
})],
};
Step 7 — Custom Integrations Wrap internal SDK calls with spans for tracing visibility:
import * as Sentry from '@sentry/node' ;
export function withSpan<T>(op : string , desc : string , fn : () => Promise <T>, attrs ?: Record <string , string | number >): Promise <T> {
return Sentry .startSpan ({ name : desc, op, attributes : attrs }, async (span) => {
try { const r = await fn (); span.setStatus ({ code : 1 , message : 'ok' }); return r; }
catch (e) { span.setStatus ({ code : 2 , message : String (e) }); throw e; }
});
}
Step 8 — SentryService Wrapper Production wrapper with singleton, metrics, and graceful shutdown:
import * as Sentry from '@sentry/node' ;
export class SentryService {
private static instance : SentryService | null = null ;
private initialized = false ;
private constructor (private config : { serviceName: string ; dsn: string ; version?: string }static getInstance (config ?: { serviceName : string ; dsn : string ; version ?: string }): SentryService {
if (!SentryService .instance ) {
if (!config) throw new Error ('Config required on first call' );
SentryService .instance = new SentryService (config);
}
return SentryService .instance ;
}
init (): void {
if (this .initialized ) return ;
const { initSentry } = require ('./sentry' );
initSentry (this .config );
this .initialized = true ;
}
captureError (error : Error , ctx ?: { tags ?: Record <string , string >; extra ?: Record <string , unknown >; level ?: Sentry .SeverityLevel }): string {
return Sentry .withScope ((scope ) => {
if (ctx?.tags ) Object .entries (ctx.tags ).forEach (([k, v] ) => scope.setTag (k, v));
if (ctx?.extra ) Object .entries (ctx.extra ).forEach (([k, v] ) => scope.setExtra (k, v));
if (ctx?.level ) scope.setLevel (ctx.level );
return Sentry .captureException (error);
});
}
async trackOperation<T>(name : string , op : string , fn : () => Promise <T>): Promise <T> {
return Sentry .startSpan ({ name, op }, async (span) => {
try { const r = await fn (); span.setStatus ({ code : 1 , message : 'ok' }); return r; }
catch (e) { span.setStatus ({ code : 2 , message : String (e) }); Sentry .captureException (e); throw e; }
});
}
async shutdown (timeoutMs = 5000 ): Promise <void > {
await Sentry .close (timeoutMs);
SentryService .instance = null ;
this .initialized = false ;
}
}
Step 9 — Health Check Exclusion Define health routes BEFORE Sentry middleware so they never create transactions:
app.get ('/health' , (_req, res ) => res.status (200 ).json ({ status : 'ok' }));
app.get ('/readiness' , async (_req, res) => {
const dbOk = await checkDatabase ();
res.status (dbOk ? 200 : 503 ).json ({ db : dbOk });
});
Output
Centralized lib/sentry.ts enforcing PII scrubbing, sample rates, and noise filters across all services
Project-per-service topology with team ownership, independent quotas, and per-service releases
Error middleware for Express and FastAPI with severity-based capture
Distributed tracing across HTTP (automatic) and message queues (manual propagation helpers)
Team-based alert routing with 4-tier escalation (P0-P3)
Source map pipeline with CLI and Webpack plugin for monorepo builds
SentryService singleton with custom metrics and graceful shutdown
Error Handling Error Cause Solution Traces not linking cross-service Missing trace headers in non-HTTP transport Use publishWithTrace/consumeWithTrace for queues init() called multiple timesMultiple imports Use SentryService singleton (idempotent init) Source maps not resolving Wrong url-prefix Use --url-prefix "~/" and --validate flag Alerts routing to wrong team Ownership rules mismatch Verify path: rules match source tree; add tags.service: fallback Health checks consuming quota Probes hitting instrumented routes Define health routes before middleware; use tracesSampler PII leaking Missing beforeSend scrubbing Enforce sendDefaultPii: false + header deletion in shared config
Examples Bootstrap a service (3 lines):
import { SentryService } from '@acme/sentry-config' ;
const sentry = SentryService .getInstance ({ serviceName : 'user-service' , dsn : process.env .SENTRY_DSN ! });
sentry.init ();
Capture with business context:
sentry.captureError (err, {
tags : { 'payment.provider' : 'stripe' },
extra : { orderId : order.id , amount : order.total },
level : 'error' ,
});
const result = await sentry.trackOperation ('order.fulfillment' , 'business.process' , () => fulfillOrder (id));
Resources
Next Steps
Roll out incrementally — start with one high-traffic service, validate traces and alerts, then onboard remaining services
Set up Sentry Crons for scheduled job monitoring (ETL, billing workers)
Enable Session Replay (@sentry/browser) for frontend error-to-session correlation
Define per-project quota budgets to prevent one noisy service from exhausting org quota
Build Discover dashboards for cross-service error trends (count() by service, level)
