Observability built for
the AI-native developer
Rectifyr unifies logs, metrics, traces, and RUM in one platform. GitHub Copilot, Claude, Openclaw, Ollama, LM Studio, and OpenAI Codex scaffold and configure your entire telemetry stack — automatically.
Opensource · Self-hostable · No vendor lock-in
What is Rectifyr?
Rectifyr is a stylized spelling of "rectifier" — a device that converts AC to DC power. It also plays on the verb "rectify," meaning to correct or put right.
The name draws inspiration from Latin: emendo (to correct and repair) and praesidium (protection or guard). Together, they suggest a tool that rectifies problems and protects systems.
Rectifyr is your DevOps Copilot.
- Converts chaos into clarity — transforms raw telemetry into actionable insights
- Corrects and protects — identifies issues before they impact users
- AI-native intelligence — powered by AI to help you understand and fix problems faster
Start Rectifyr with Docker Compose
Prefer to run Rectifyr directly? These upstream commands launch a local self-hosted instance from the core SOBS repository.
- Use the official upstream SOBS image
- Run either Docker or docker-compose quick start
- Open the Rectifyr UI on localhost
# Docker
docker run -p 44317:4317 -v sobs_data:/data ghcr.io/abartrim/sobs:latest
# docker-compose
docker-compose up -d
# Open local Rectifyr dashboard
open http://localhost:44317
A quick look inside Rectifyr
Real views from the Rectifyr interface (captured from upstream SOBS), including logs, traces, dashboards, quick tour flows, and AI-powered analysis workflows.
Quick Tour: Summary and Signal Health
Agent Flow and AI Transparency
Dashboard Overview
Logs and Search
Distributed Tracing
Trace Drilldown
Natural-Language Query
AI Assistant Summary
From zero to instrumented
in under 60 seconds
Ask GitHub Copilot, Claude, Openclaw, or any MCP-compatible AI assistant to set up Rectifyr for your application. The AI reads your codebase, picks the right SDK, and generates a ready-to-run telemetry configuration — no manual YAML required.
- Auto-detects your language and framework
- Generates OTel instrumentation code
- Creates dashboards and alert rules automatically
- Works with self-hosted LLM tools (Ollama, LM Studio, vLLM)
# Ask Copilot in any terminal or IDE:
# "@rectifyr scaffold Python FastAPI telemetry"
from opentelemetry import trace, metrics
from opentelemetry.exporter.otlp.proto.grpc.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
# Auto-generated by Rectifyr AI onboarding
provider = TracerProvider()
provider.add_span_processor(
BatchSpanProcessor(
OTLPSpanExporter(
endpoint="https://ingest.rectifyr.io:4317",
headers={"x-rectifyr-token": "<YOUR_TOKEN>"}
)
)
)
trace.set_tracer_provider(provider)
tracer = trace.get_tracer("my-fastapi-app")
# Your app is now sending traces to Rectifyr
From setup to telemetry in minutes
A short guided setup flow from onboarding to first telemetry signal.
Everything you need to understand
your system
Industry-standard signals — logs, metrics, traces, profiles, and real user monitoring — unified in one Opensource platform.
Logs, Metrics & Traces
Full OpenTelemetry support via gRPC and HTTP. Ingest from any language or framework using standard OTel SDKs.
Real User Monitoring
Capture front-end performance, Core Web Vitals, session replay, and user journeys with a lightweight JS snippet.
Alerts & Incidents
Multi-condition alerts with PagerDuty, Slack, and webhook delivery. Auto-grouping reduces notification fatigue.
AI Anomaly Detection
ML-powered baseline learning flags regressions and anomalies before they become outages. No thresholds to set.
SSO & Multi-tenancy
Rectifyr login with OIDC / SAML authentication. Isolate teams and projects with fine-grained RBAC.
ClickHouse Storage
Petabyte-scale columnar storage. Sub-second queries over billions of log events. Compressed by default.
Monitoring that
operates itself
Automated seasonal monitoring adjusts baselines as your traffic patterns evolve — no manual tuning. When issues are detected, Rectifyr opens GitHub Issues with AI-proposed fixes and waits for your approval before applying them.
Auto-adjusts thresholds for Black Friday, end-of-month, or any recurring traffic pattern.
Copilot or Claude drafts a code fix, posts it as a PR, and waits for engineer approval.
Human-in-the-loop controls before any automated change is applied to production.
Root-cause analysis written in plain English, cited directly from your traces and logs.
Anomaly detected: /api/checkout P99 latency jumped from 120ms 4.2s at 14:32 UTC.
Root cause (AI analysis): Trace sampling shows 94% of slow requests waiting on PaymentServiceClient.charge() — upstream circuit breaker tripped.
Proposed fix (Copilot PR #847):
+ client = PaymentServiceClient(retry=retry_policy)
Works with your AI coding assistant
Rectifyr exposes a Model Context Protocol (MCP) server so any AI assistant can query your observability data directly in context.
GitHub Copilot MCP
Ask Copilot: "Why is my checkout service slow?" — it queries Rectifyr traces and answers with specific span-level context right in your IDE.
Claude Code
Use claude code to scaffold a complete observability
setup, generate runbooks from your alert history, or perform
natural-language log searches.
OpenAI Codex
Codex-powered agents can create and maintain observability configurations as part of your CI/CD pipeline — keeping your instrumentation in sync with your code.
Also works with Openclaw, Cursor, Windsurf, Cline, Continue, LM Studio, Ollama workflows, and any MCP-compatible tool
Get started — it's freeStart free. Scale as you grow.
A generous free tier for individuals. Affordable per-space + per-user pricing for teams. Enterprise plans with dedicated infrastructure.
Ready to ship better software?
Join developers and teams using Rectifyr to understand their systems — powered by AI, built on open source.