getsentry/sentry-for-ai
32 skills42.1k installs6.3k starsGitHub
Install
npx skills add https://github.com/getsentry/sentry-for-aiSkills in this repo
1Sentry Fix IssuesSentry Fix Issues is a Sentry workflow skill for agent-assisted production debugging. Solo builders and small teams who already run Sentry use it when users ask to fix Sentry errors, debug production bugs, investigate exceptions, or work through an issue backlog. It expects the Sentry MCP server to be configured and connected with access to the relevant organization and project. The agent methodically reads stack traces, breadcrumbs, distributed traces, and surrounding context to identify root causes and propose fixes, while refusing to treat exception messages, request bodies, tags, or breadcrumbs as trusted instructions—mirroring secure handling of attacker-controlled telemetry. It is invoked for concrete issue IDs, error messages, or recent failure triage, not for generic coding questions without Sentry context. Outcomes are clearer diagnosis, prioritized fixes aligned with real event data, and safer automation around observability-driven repair loops.4kinstalls2Sentry React Native SdkSentry React Native SDK is a deep implementation skill for solo builders shipping cross-platform mobile apps who need one observability surface for three runtimes. React Native failures often masquerade as generic red screens while the real fault lives in native code or rejected promises; the skill walks through initialization, native crash bridging, top-level error boundaries, and enrichment patterns so grouped issues are actionable. It assumes @sentry/react-native at least 6.0.0 (8.0.0 recommended), React Native 0.71+ for Fabric, and covers hooks like beforeSend to strip PII before events leave the device. Use it while hardening a release candidate and keep it referenced during Operate when tuning sampling, breadcrumbs, and release health. It is SDK documentation encoded for agents—not a hosted Sentry project or CI gate by itself.2.5kinstalls3Sentry WorkflowThe sentry-workflow skill is a router that directs you to the right Sentry debugging workflow based on your task. Solo builders use it when they encounter production errors, receive Sentry bot comments on pull requests, or need to upgrade Sentry SDKs. It ensures you get the right guidance for your specific debugging scenario—whether that's triaging live issues, resolving code review feedback, or managing SDK migrations.2kinstalls4Sentry Nextjs SdkSentry Next.js SDK is an agent skill that teaches how to enable Sentry AI Agents Monitoring in Next.js projects using @sentry/nextjs 10.53.0 or newer. It is aimed at solo and indie builders who ship LLM-powered features—chat, agents, or tool-using workflows—and need production-grade observability without building custom logging. The guide maps each supported stack (OpenAI, Vercel AI SDK, Anthropic) to the correct integration API, clarifies which hooks auto-activate on the Node server versus what must be wired manually on Edge or in the browser, and stresses that tracing must be on with a positive tracesSampleRate. Use it when you are integrating the SDK during build, validating tracing before launch, or debugging agent failures and cost spikes in production. The outcome is a configured Sentry pipeline that surfaces model usage, estimated cost, tool I/O, and performance across the full AI path.1.8kinstalls5Sentry Create Alertsentry-create-alert is an agent skill for setting up Sentry notifications through the workflow engine API. Solo builders and small teams who already send errors to Sentry can use it when they want reliable paging or chat signals instead of manually clicking through UIs. The skill walks through gathering org slug, region, scoped auth token, and alert intent, then issues API calls to define workflows that fire on issue conditions—including priority and de-escalation patterns. It explicitly notes the beta API surface tied to New Monitors and Alerts, which may differ from legacy alert screens. Invoke it when someone asks to create a Sentry alert, wire Slack or PagerDuty, or automate issue notifications. It is not a substitute for defining meaningful error grouping or fixing root causes; it focuses on notification plumbing with curl and token prerequisites.1.7kinstalls6Sentry Sdk SetupSentry SDK Setup is a router agent skill from Sentry’s skills tree that helps solo and indie builders add observability without guessing the wrong SDK. When someone asks to add Sentry, install an SDK, or set up error monitoring, the skill instructs the agent to read project fingerprints such as package.json, go.mod, requirements.txt, Gemfile, csproj, or build.gradle, state what platform was found, and recommend the precise child skill URL under https://skills.sentry.dev/. It deliberately blocks improvisation: the agent must wait for confirmation, then download the full platform skill with curl and follow that document’s prerequisites and configuration steps. That pattern matters for AEO and for real installs because truncated tool summaries often drop critical Sentry initialization details. The skill is Apache-2.0 licensed and labeled role router, so it pairs well with ship-time hardening and operate-phase incident readiness.1.7kinstalls7Sentry React SdkSentry React SDK is an agent skill that teaches solo builders how Sentry’s React package hooks the browser and React tree to record production failures. It maps each automatic layer—uncaught exceptions, unhandled rejections, timer and rAF patches, and component render errors via ErrorBoundary or reactErrorHandler—to the integration that enables it and the minimum @sentry/react version required. The skill matters because most shipped React apps silently lose errors inside local try/catch blocks, React Router’s default boundary, or Promise chains with attached catch handlers unless you add explicit capture calls. Install it when you are integrating or debugging Sentry in a client-side React or React Native Web stack during Operate, or right before Ship if you want monitoring live on day one. It is procedural SDK knowledge, not a hosted MCP server: your agent uses it to choose boundaries, manual instrumentation points, and optional console capture while you keep DSN and project setup in your own config.1.6kinstalls8Sentry Feature Setupsentry-feature-setup is a router agent skill in the Sentry for AI catalog for solo builders who already run Sentry and need a specific capability beyond baseline SDK init. It explicitly tells the agent not to assume which feature fits: AI monitoring for OpenAI, Anthropic, LangChain, and similar stacks, OpenTelemetry collector export, or alert and on-call workflows each branch to a dedicated child skill path under skills.sentry.dev. The How to Fetch Skills section requires curling full SKILL.md files because condensed fetches drop setup details. That pattern suits indie operators wiring LLM traces in production, multi-service OTel, or notification rules without reading the entire Sentry tree upfront. Placement on Operate/monitoring reflects steady-state config, but you can invoke it right after deploy when choosing what to instrument next.1.6kinstalls9Sentry Node Sdksentry-node-sdk is Sentry’s guided setup skill for server-side JavaScript and TypeScript on Node.js, Bun, and Deno. It walks a solo builder through installing the right package, early instrumentation entrypoints, and optional products—distributed tracing, structured logging, profiling, cron monitors, metrics, and AI monitoring—without guessing framework-specific boot order. The wizard mentality matches indie API and worker projects where one developer owns both the repo and the on-call pager. It explicitly defers NestJS to sentry-nestjs-sdk and Next.js to sentry-nextjs-sdk so agents do not apply the wrong initializer. Builders typically invoke it right before or right after first deploy when “add Sentry” means more than dropping a DSN snippet: preload hooks, release health, and framework middleware must align or events never arrive.1.6kinstalls10Sentry Code Reviewsentry-code-review is a workflow skill for developers who already run Sentry on their app and see automated bug findings on GitHub pull requests. When someone asks to “fix what Sentry flagged on this PR” or to sweep open PRs for sentry[bot] threads, the skill treats each comment as structured input: file path, line, severity, confidence, and markdown bodies with HTML details blocks. It is built for solo and small-team shippers who cannot afford to ignore CRITICAL annotations but also do not want to manually decode every collapsible analysis section. Typical flow: locate the PR (by number or recent search), map comments to local files, propose or apply edits, and validate that the reported failure mode is addressed. It pairs naturally with the parent sentry-workflow catalog entry and assumes GitHub is the review surface. It is not a substitute for full human security review or for configuring Sentry projects—that remains upstream—but it compresses the loop from bot comment to patched diff.1.5kinstalls11Sentry Sdk UpgradeSentry SDK Upgrade is an agent skill for solo and indie builders who already ship JavaScript or TypeScript apps with Sentry and need to move to a newer major SDK without leaving stale configs behind. It walks through systematic discovery: ripgrep-style searches for @sentry imports, locating sentry and instrumentation config files, and comparing package.json entries to installed versions under node_modules. A seven-framework matrix maps where client, server, build, and edge initialization typically lives for Next.js, Nuxt, SvelteKit, Remix, React SPAs, Angular, and Vue, so upgrades touch every entry point instead of only the obvious file. Use it during dependency maintenance sprints, after Sentry migration guides change defaults, or when a major bump warns about removed APIs. The skill packages procedural shell commands your coding agent can run in-repo; it is not a hosted MCP server and does not replace reading official Sentry changelog notes for your target version.1.5kinstalls12Sentry Setup Ai MonitoringSentry Setup AI Monitoring teaches solo builders how to keep AI agent work visible in Sentry when default performance sampling would silently discard whole traces. Agent runs are span trees: if the root transaction is dropped, every child gen_ai span disappears—so a global 20% rate can mean zero captured agent runs. The skill documents two architectures: standalone gen_ai roots (background jobs, queues, CLI) where the sampler matches gen_ai attributes directly, and the common web pattern where POST /api/chat is the root and must be sampled at full rate before any model code executes. It provides ready-to-adapt JavaScript and Python init snippets using tracesSampler or traces_sampler, attribute checks for sentry.op and gen_ai.system, and guidance to tune non-AI baseline sampling separately. Invoke while wiring observability for LLM features you already ship or are about to launch.1.5kinstalls13Sentry Browser SdkSentry Browser SDK is an agent skill that teaches solo builders how client-side error monitoring actually works in the browser: which failures are captured automatically through window handlers and patched APIs, and which require explicit Sentry.captureException or captureMessage calls. It is aimed at indie developers shipping SPAs, marketing sites, or browser extensions who need trustworthy crash data without building a custom logging pipeline. Use it while integrating or hardening observability after launch, when debugging production-only issues, or when reviewing whether your try/catch blocks are hiding incidents from Sentry. The skill matters because silent failures in user browsers directly affect churn and support load; aligning SDK defaults with manual instrumentation prevents false confidence that “Sentry is installed” means every business failure is tracked.1.5kinstalls14Sentry Pr Code Reviewsentry-pr-code-review is a workflow skill in the Sentry-for-AI tree that helps solo builders and small teams close the loop on automated PR feedback from Seer Bug Prediction. When someone asks to address Sentry review comments or shares a PR number, the agent follows documented GitHub CLI steps to list inline comments authored by seer-by-sentry[bot], inspect file and line context, and apply code fixes aligned with each finding. It also supports discovery when no PR is specified—finding recent pull requests that still have unresolved bot threads. The skill explicitly warns that comment formatting is not a stable API contract, so parsers must be verified against live comments. It depends on authenticated gh and the Seer GitHub App installation; it does not replace human security review or full CI. Parent linkage is sentry-workflow for broader Sentry agent operations.1.4kinstalls15Sentry Cocoa SdkSentry Cocoa SDK is a reference skill for solo and indie builders shipping Apple-platform apps who need production-grade error monitoring without reading the entire Sentry docs. It maps the master switches and tuning options you actually set in code: crash reporting, sampling, breadcrumbs, app-hang detection with timeout intervals, watchdog termination tracking, and automatic capture of failed HTTP requests against configurable status ranges and host patterns. Use it when you are integrating Sentry into a new Swift or Objective-C project, debugging why events are missing or noisy, or aligning agent-generated init code with current SDK defaults. The skill is integration-focused—not a generic debugging methodology—so it pairs naturally with ship-phase testing and operate-phase on-call workflows once traffic is live.1.3kinstalls16Sentry Python SdkSentry Python SDK is a structured setup skill for solo builders adding production-grade observability to Python applications. It belongs to the broader Sentry SDK setup tree and is invoked when someone asks to add Sentry, install sentry-sdk, or configure monitoring across common async and sync web stacks plus Celery workers. Phase one emphasizes detection—scanning dependency manifests and inferring frameworks—so recommendations match what is already in the repo rather than generic snippets. Subsequent guidance covers the full Sentry surface area relevant to shipping: exceptions, performance traces, profiles, structured logs, metrics, cron monitors, and AI-related instrumentation where applicable. The skill is opinionated but defers to current Sentry documentation for version-specific APIs, which reduces drift for indies who ship frequently but do not maintain an internal observability playbook.1.2kinstalls17Sentry Sdk Skill CreatorSentry SDK Skill Creator documents how to build complete, opinionated Sentry SDK skill bundles for coding agents. Solo and indie builders use it when they want one installable skill that walks an agent through error monitoring, distributed tracing, profiling, structured logging, metrics, cron monitoring, and frontend session replay—without bloating the main SKILL.md. The canonical pattern is a wizard entry file that stays short while references/ holds deep-dive setup steps loaded on demand. That matches how real projects adopt observability incrementally: start with errors, add tracing when services talk to each other, turn on profiling or logs when debugging gets painful. The skill is meta: it teaches bundle layout, naming (sentry-<platform>-sdk), and how agents should act as experts who read the repo and recommend pillars. It fits builders shipping SaaS, APIs, CLIs, and agent-backed apps who publish skills to Claude Code, Cursor, or Codex ecosystems and want Sentry setup to feel like a guided wizard rather than scattered docs.1.2kinstalls18Sentry Nestjs SdkSentry NestJS SDK is an integration-focused agent skill for solo builders running Node APIs on NestJS who need production-grade observability without reading the entire Sentry docs tree first. It activates when you ask to add Sentry, install @sentry/nestjs, or configure error monitoring, performance tracing, profiling, structured logging, metrics, cron monitoring, or AI monitoring in NestJS apps spanning Express or Fastify, GraphQL resolvers, microservices, WebSockets, and background jobs. The workflow begins with explicit detection commands against package.json and runtime versions, then walks through opinionated setup aligned with @sentry/nestjs 10.x (NestJS 8–11). It sits under the parent sentry-sdk-setup skill so patterns stay consistent with other framework guides. Use it during backend hardening before launch or when incidents prove you need centralized stack traces and trace context in production.1.2kinstalls19Sentry Otel Exporter SetupThis skill guides you through configuring the OpenTelemetry Collector to send traces and logs to Sentry using the Sentry Exporter component. Use it when you're setting up observability across multiple services, need to route telemetry to different Sentry projects, or want automatic project creation based on incoming data. It matters because it unifies your telemetry infrastructure and provides a single entrypoint for traces and logs from distributed systems.1.1kinstalls20Sentry Dotnet SdkSentry .NET SDK skill gives solo builders copy-paste guidance for Sentry Cron Monitoring on .NET services and scheduled work. It explains when to emit CheckInStatus.InProgress versus Ok or Error, how to correlate start and end signals with a saved checkInId, and when a one-shot heartbeat is enough versus the recommended two-signal flow that catches both missed schedules and hung jobs. The material is aimed at indie teams shipping APIs or SaaS backends on .NET who already use Sentry for errors and want the same project to page when a nightly job never starts or dies mid-run. Invoke it while hardening Operate workflows—after deploy—not during greenfield feature design. It is reference-style integration knowledge rather than a generic debugger; pair it with your existing Sentry DSN setup and monitor slug naming in the Sentry UI.1.1kinstalls21Sentry Android SdkSentry Android SDK (Crons / Monitors) is a narrow, high-signal agent skill for indie mobile developers who already ship with Sentry and need scheduled-job visibility on Android. The skill explains that Crons use the core Java SDK—there is no first-party WorkManager or AlarmManager adapter—so you implement check-ins explicitly around sync workers, nightly cleanup, or subscription refresh tasks. It contrasts the experimental CheckInUtils helper with manual captureCheckIn flows suited to production stability requirements. Solo builders use it in Operate when crashes alone are insufficient and silent job failures would corrupt data or billing state. The content aligns with official Java Crons documentation where Android-specific pages may be missing, reducing copy-paste errors on status enums and monitor slugs. Expect integration guidance, not a generated Gradle module.1.1kinstalls22Sentry Php Sdksentry-php-sdk is an agent skill that teaches how to connect PHP and Laravel scheduled work to Sentry Crons using check-ins at job start, completion, and failure. Solo builders shipping cron scripts, queue workers, or Laravel scheduler entries use it when they need alerts when jobs slip, crash, or never run—not just when PHP throws an uncaught exception elsewhere. The skill contrasts the simplest withMonitor() callable wrapper, full manual captureCheckIn() control for custom timing and status, and the Laravel macro for scheduled tasks with minimal boilerplate. It also explains heartbeat check-ins for “job didn’t start” scenarios versus runtime failures. The audience is indie SaaS and API operators already on Sentry who want cron visibility without building a custom watchdog.1.1kinstalls23Sentry Go SdkThe sentry-go-sdk skill documents how to instrument Go programs with Sentry Cron Monitoring so scheduled work is observable in production. Solo builders shipping Go APIs, workers, or CLI cron wrappers often ship the happy path first and only learn a job stopped running when data goes stale. This skill explains Init with DSN, async check-ins, MonitorSlug alignment with Sentry UI, and the three CheckInStatus values that distinguish healthy completion from failure. It covers grace margins before a run counts as missed and max runtime before an in-progress check-in times out. Minimum SDK v0.18.0 applies. Use it while hardening Ship/Operate handoff for any Go binary that runs on a schedule—CI deploys, VPS cron, or Kubernetes CronJob—when you already use Sentry for errors and want one pane for cron health.1kinstalls24Sentry Flutter SdkSentry Flutter SDK is an agent skill that walks solo and indie builders through end-to-end Sentry setup for Flutter and Dart applications. Invoke it when someone asks to add Sentry, install sentry_flutter, or configure error monitoring, tracing, profiling, session replay, or logging across mobile and desktop targets. The workflow opens with a Detect phase—shell commands to infer project shape—before prescribing opinionated configuration, so you avoid half-wired DSNs or missing native crash capture. It aligns with current stable APIs (documented as sentry_flutter ≥9.14.0 as of February 2026) while reminding you to verify against Sentry’s Flutter documentation before merging. For a builder shipping a cross-platform app alone, this compresses days of observability spelunking into a repeatable agent run tied to real invoke phrases like “setup Sentry in Dart” or “monitor native crashes.” Parent skill tree context (SDK Setup) signals it is one concrete platform package under Sentry’s broader setup family, not a generic logging snippet.1kinstalls25Sentry Svelte SdkSentry Svelte SDK is an agent skill for solo builders shipping Svelte or SvelteKit apps who need trustworthy error telemetry without reading every integration doc first. It explains how automatic capture works across the browser (unhandled errors, promise rejections, timer callbacks) and the SvelteKit server and client hooks, including `handleErrorWithSentry` and `sentryHandle` for load functions, route handlers, and navigation failures. You use it during Operate when you are adding or hardening monitoring, or late in Build when the stack is SvelteKit and you want hooks wired correctly before launch. The write-up emphasizes that no extra configuration is required beyond initialization for baseline capture, while still showing how to compose a custom error handler after Sentry. For indie teams on Vite env vars such as `VITE_SENTRY_DSN`, it is a focused integration reference rather than a generic debugging methodology.987installs26Sentry Cloudflare SdkSentry Cloudflare SDK (Crons / Job Monitoring) is an integration-oriented agent skill from Sentry’s sentry-for-ai repo that helps solo builders instrument Cloudflare Workers scheduled tasks with Sentry Crons. It explains how missed check-in margins, exceeded maxRuntime, and explicit error statuses surface as alerts—so indie operators catch silent cron failures before data pipelines or cleanup jobs rot. When you wrap handlers with withSentry, the scheduled entry point gains faas.cron spans carrying the cron expression, ISO scheduled time, and timer trigger metadata, while errors propagate into Sentry without bespoke boilerplate. The skill targets TypeScript Workers using wrangler.toml cron triggers and minimum SDK versions for captureCheckIn and withMonitor. Use it during Operate when moving from “cron runs on my machine” to production schedules on Cloudflare’s edge. Intermediate complexity: you need a Sentry project, DSN in env, and basic Workers export shapes.982installs27Sentry Ruby SdkSentry Ruby SDK is an agent skill that teaches solo builders how to implement Sentry Crons monitoring in Ruby apps so scheduled jobs do not fail silently. The readme covers manual check-ins for Clockwork, Whenever, or custom loops; ActiveJob monitor modules; Sidekiq-Cron hooks; upserting monitor configuration; and completion-only check-ins when start events are impractical. You use it when production depends on daily reports, billing runs, or queue sweeps and you need Sentry to alert on missed schedules or slow executions. The skill is integration-focused procedural knowledge for the official SDK—not a replacement for Sentry project setup or DSN configuration. It assumes you already ship a Rails or Ruby worker stack and want copy-paste patterns that pair `:in_progress` with `:ok` or `:error` while capturing exceptions on failure.970installs28Sentry Elixir Sdksentry-elixir-sdk (Crons) guides solo builders and small Elixir teams to add Sentry Cron Monitoring so scheduled work cannot fail quietly. The SDK supports manual check-ins from GenServer or custom loops, plus integrations aligned with Oban and Quantum. The documented pattern signals in_progress at job start, then ok or error when work finishes, optionally creating monitors via upsert config so you do not hand-configure every slug in the Sentry UI. Use this when shipping Phoenix or API backends that rely on nightly reports, queue consumers, or cron-style maintenance. It sits squarely in Operate because value appears after deploy, when you need proof that periodic jobs still run and complete within expected time.893installs29Sentry React Router Framework SdkSentry React Router Framework SDK is an agent skill wizard for solo builders shipping React Router v7 framework-mode apps who need structured Sentry installation instead of copy-pasting snippets. It walks through project detection, then guides client and server initialization for error monitoring, performance tracing, profiling, session replay, structured logs, and user feedback when users ask to add Sentry or configure @sentry/react-router. The skill targets apps using framework entry files and React Router–specific integrations, and it flags that the package is beta while pointing non-framework v5/v6/v7 setups to the separate React SDK skill. For indie teams wearing dev and ops hats, it reduces missed server-side failures and inconsistent client/server DSN setup during the stretch between feature-complete and real users.633installs30Sentry Tanstack Start SdkSentry TanStack Start React SDK is an agent skill that teaches solo builders how to integrate @sentry/tanstackstart-react into TanStack Start projects for end-to-end error visibility. It walks through the minimum alpha SDK version, where the framework auto-captures unhandled client exceptions, promise rejections, server request failures, and server function failures once global middleware and fetch wrapping are in place. The skill stresses a practical rule many integrations omit: anything caught without rethrow must be reported manually, including SSR render failures and errors absorbed by React error boundaries. Copy-paste patterns cover src/start.ts middleware registration and the server entry wrapper so agents do not guess at TanStack Start’s split client/server model. For indie SaaS teams shipping on TanStack Start RC, this skill reduces time-to-first-actionable-error-event compared to ad-hoc console logging and makes agent-assisted refactors safer because monitoring hooks stay aligned with official Sentry mechanisms.614installs31Sentry Instrumentation Guidesentry-instrumentation-guide is a companion deep dive to Sentry for AI that teaches solo builders how to place observability data in the correct Sentry product surface. Instead of dumping logs or emitting oversized custom events, you learn when an exception should open an Issue, when a slow LLM tool call belongs in a trace span, and how metrics and logs backfill gaps where no stack trace fired. The narrative covers workflow differences—errors drive assignment and regression alerts, while successful empty queries are not errors even if users are unhappy—and walks through retention and overlap rules across signals. Indie developers shipping agent APIs or multi-step inference pipelines can use it while wiring instrumentation in Ship and while triaging Operate incidents, reducing alert fatigue and blind spots. It expects familiarity with Sentry concepts from the parent SKILL.md decision table and extends that with reasoning, gotchas, and anti-patterns for derive-everything-from-one-event shortcuts.1installs32Sentry Span StreamingSentry Span Streaming is a feature-setup agent skill for solo and indie builders who already use Sentry and want traces to leave the app incrementally instead of waiting for a finished transaction envelope. It walks through enabling span-first trace lifecycle, wiring span streaming integration, and validating that your JavaScript stack (browser, Node, Bun, Deno, or Cloudflare) supports the streamed model today. The skill is narrow by design: it does not pretend Python, Ruby, Go, or other SDKs are ready when the guide marks them as not yet available. You reach for it when latency, partial failure visibility, or downstream per-span processing matters more than legacy transaction-centric batching. Parent context from sentry-feature-setup keeps the migration aligned with the rest of your Sentry feature rollout rather than a one-off config tweak.1installs