Laptop-native observability for LLMs, microservices, and APIs. Zero-config setup. Privacy-first. Offline-capable. Ship with confidence.
macOS, Linux, Windows. 5-minute setup. No signup required.
You ship to staging. Then you find out what broke. Sound familiar?
You cannot observe LLM calls, microservices, or API requests locally. “Works on my machine” is a guess, not a fact. Your local dev environment gives you zero visibility into what your code actually does at runtime.
“I can’t see what my LLM calls are doing until I ship to staging.”
See how DevScope solves thisDeploy to staging. Wait. Check logs. Find the issue. Fix it. Redeploy. That is your feedback loop — 30+ minutes for a 2-line change. You are spending more time waiting than building.
“I spend 20 minutes deploying to staging just to test a 2-line change.”
See 5-minute setupCloud-based APM sends your telemetry offsite. Your LLM traces include sensitive prompts, customer data, PII. Compliance blocked your observability tool request. You are stuck without visibility because your current options cannot stay local.
“Privacy matters — I don’t want my LLM prompts sent to third-party APM vendors.”
See privacy architectureLocal observability that does not slow you down.
All traces stay on your laptop. No cloud calls. No third-party APIs. No data exfiltration. DevScope runs a local collector and dashboard — your telemetry never touches an external server.
Compliance-friendly for sensitive workloads. Your LLM prompts, customer data, and PII stay where they belong — on your machine. No vendor access. No audit headaches.
“Monitor LLM calls to Claude and GPT without sending traces to third-party APM. Compliance approved it in one day.”
Async instrumentation adds less than 50ms to your request lifecycle. DevScope hooks into your framework at the middleware layer and collects traces without blocking your application threads.
Your dev loop stays fast. No perceptible slowdown. No “observability tax” on local performance. You get visibility without sacrificing the speed you depend on.
“Add observability to a local Flask API without adding 200ms to every request. Most developers do not notice any slowdown.”
Install via CLI. DevScope auto-instruments Flask, FastAPI, Django, Express, Next.js, and more. Zero configuration required. No YAML files. No environment variables. No code changes.
You get value in 5 minutes, not 5 hours. Install, run your app, and see traces. That is the entire setup. No day-long configuration rituals. No documentation deep-dives before first value.
“Install DevScope, run your app, see traces — all in 5 minutes. Zero config. No code changes.”
Observe OpenAI, Anthropic, and Ollama calls. Trace Docker containers, local APIs, Redis, Postgres. See the full request path — from your API gateway through your LLM provider and back — all on localhost.
Modern dev stacks are not just HTTP requests. You are building with LLMs, running microservices in Docker, querying multiple databases. DevScope sees all of it. Not just the web layer.
“See LLM prompt, retrieval, response flow for local RAG development. Track cost per call, latency per provider, tokens per request.”
Zero-config default. Auto-instrumentation. Offline-capable. Total time: 5 minutes.
One command. All platforms. DevScope installs a local collector, dashboard server, and auto-instrumentation hooks. No dependencies. No Docker required for the tool itself.
curl -sSL https://devscope.sh/install.sh | bashSupports macOS (ARM + Intel), Linux (x86_64, ARM64), Windows (WSL2).
Start your application as usual. DevScope auto-instruments Flask, FastAPI, Django, Express, Next.js, OpenAI SDK, Anthropic SDK, LangChain, Redis, and Postgres. No decorators. No middleware registration. No config files.
devscope run -- python app.pyAuto-instruments 15+ frameworks. Manual instrumentation available via OpenTelemetry SDK for unsupported stacks.
See Supported FrameworksNavigate to localhost:7878 in your browser. See traces, metrics, and service topology. Filter by LLM calls, API requests, database queries. Drill into individual spans. View cost, latency, and token usage for every LLM call.
open http://localhost:7878Dashboard runs locally. No internet required. No cloud login. Works offline.
Download FreeReal dashboard. Real traces. Real developer productivity.
Join 5K+ engineers shipping with confidence.
“DevScope cut my local dev feedback loop from 20 minutes to 2 minutes. I see LLM calls, API latency, DB queries — all without leaving my laptop. I do not know how I built without this.”
“Privacy-first observability was non-negotiable for us. DevScope runs locally, zero data exfiltration. Compliance approved it in one day. That never happens with observability tools.”
“My team adopted DevScope in one sprint. Zero-config setup. Works with our stack — FastAPI, Redis, Postgres, OpenAI. Now every engineer observes their code before merging. Release confidence went through the roof.”
MIT License. 5K+ downloads. 100+ GitHub stars. Built in public.
Star the repo. Fork it. Contribute. The code is the proof.
Team dashboards. Persistent storage. 24/7 support. Start free. Upgrade when your team needs it.
Shared observability across 5-50 engineers. Cross-team trace aggregation. See what your entire engineering org is observing — not just your own laptop.
$100/month per team (5-user limit)Retain traces for 90 days (vs. 14-day local storage on free tier). Export to Elasticsearch or S3 for long-term analysis. Audit trails for compliance teams.
Included in Team tierRun the DevScope server on your own infrastructure. Air-gapped environments. Data residency compliance. Your telemetry stays in your data center.
Custom pricingSlack and Teams integration. Dedicated support engineer. 24-hour response SLA. Priority bug fixes. Direct access to the engineering team that builds DevScope.
Included in Enterprise tier24-hour response SLA. Custom deployment. Volume discounts available.
DevScope adds less than 50ms overhead via async instrumentation. The collector runs in a separate process. Your application threads are not blocked. Most developers do not notice any slowdown.
If you need even less overhead, disable auto-instrumentation and manually instrument only the code paths you care about.
View benchmark testsFive minutes end-to-end. Install via CLI (1 minute). Run your app with devscope run (2 minutes). Open your dashboard at localhost:7878 (2 minutes).
Zero configuration required for Flask, FastAPI, Django, Express, Next.js, OpenAI, Anthropic, Redis, and Postgres. Manual instrumentation is available for other frameworks via the OpenTelemetry SDK.
View quick-start guideNowhere. All traces stay on your laptop. DevScope runs a local collector and dashboard. Zero external API calls. Zero cloud dependencies. Zero data exfiltration.
Your LLM prompts, API payloads, database queries, and trace data never leave your machine. There is no phone-home, no telemetry upload, no “anonymous usage data” sent to our servers. If you are offline, DevScope works exactly the same.
View privacy architectureYes. 100% offline-capable. No internet connection required. No VPN. No cloud login. DevScope runs entirely on your local machine.
Trace your code on an airplane. Debug in a coffee shop without Wi-Fi. Work in an air-gapped environment. DevScope does not need the network for any core functionality.
View offline mode documentationAuto-instrumentation is available for:
Web Frameworks: Flask, FastAPI, Django, Express, Next.js
LLM Providers: OpenAI, Anthropic, Ollama, LangChain
Databases: PostgreSQL, MySQL, Redis, MongoDB
Infrastructure: Docker, Docker Compose
For frameworks not on this list, manual instrumentation is available via the OpenTelemetry SDK. DevScope is built on OpenTelemetry standards, so any OTel-compatible library works.
View full supported frameworks listYes. DevScope auto-instruments the OpenAI Python SDK, Anthropic Python SDK, Ollama, and LangChain. You see:
All data stays local. Your prompts are never sent to our servers or any third party.
View LLM observability guideStill have questions? Join our community Slack.
Copy the command for your platform. One line. Zero config.
macOS 12+ (Monterey, Ventura, Sonoma, Sequoia). Apple Silicon (M1/M2/M3/M4) and Intel.
curl -sSL https://devscope.sh/install.sh | bashUbuntu 20.04+, Debian 11+, Fedora 36+, CentOS Stream 9+. x86_64 and ARM64.
curl -sSL https://devscope.sh/install.sh | bashWindows 10/11 with WSL2 enabled. Ubuntu, Debian, or Fedora on WSL.
curl -sSL https://devscope.sh/install.sh | bashPrefer to download the binary manually? View GitHub Releases — for air-gapped environments, custom installations, or package manager integration.
Free forever. 5-minute setup. Privacy-first. Offline-capable.
macOS, Linux, Windows. No signup required. No credit card.