Your AI writes the code. Now let it read the logs.
Auralogs gives Cursor, Claude Code, and Codex read-only access to your real production logs over MCP, so they debug with real context instead of guessing. Stop pasting stack traces into chat.
Free forever · 50,000 logs/month · No credit card
Read-only by default
MCP/API read keys are scoped to one project and cannot mutate settings or source code.
SDKs for 8 languages
Drop-in logging for JavaScript, Python, Java, Swift, Go, Rust, C++, and Godot.
Your AI keys
AI analysis and autofix use the Anthropic or OpenAI key you configure. Auralogs does not resell inference.
GitHub stays optional
Start with structured logging and MCP. Connect a specific repo only when you want reviewed fix PRs.
Built by engineers who wanted their own AI on-call to read real logs, not screenshots. Auralogs separates ingest keys from read keys and leaves source access off until you explicitly connect GitHub.
Start with logs. Add agents later.
The first win is simple: send structured logs, copy a read key, and let your agent ask useful questions about production behavior.
Install the SDK
Use JavaScript, Python, Java, Swift, Go, Rust, C++, or Godot and attach the metadata your team already cares about.
Copy a read-only MCP/API key
New projects create scoped ingest and read keys so agents can query context without write permissions.
Ask what broke
Connect Codex, Claude, Cursor, or an internal agent and ask it to search logs, inspect traces, and summarize the incident.
Stop pasting logs into your AI tools
Give AI production context.
No copy-paste transcripts.
Connect your AI tool to the Auralogs MCP server and ask it what happened. Your agent can search logs, pull related events, inspect metadata, and summarize the incident from the same context your team sees.
Checkout failures spiking after deploy v2.3.1
Structured logging built for agent access.
Levels, search, traces, environments, and retention are the logging foundation. The difference is the surface on top: every log is queryable by humans, REST clients, and read-only AI agents through MCP.
Structured logs, five levels
Debug, info, warn, error, fatal. Attach any metadata: user IDs, traces, request context. Auralogs keeps it all queryable.
Live timeseries dashboard
See log volume by level at a glance. Zoom from the last hour to the last 90 days. Errors jump out the moment they spike.
Search, filter, trace
Full-text keyword search. Filter by level, environment, or trace ID. Follow a single user's session end-to-end.
Dev, staging, prod, separate
Tag every log with its environment. Keep noisy dev logs from drowning real prod incidents. Switch views in one click.
Two paths from the same logs
Start by sending structured logs. Use MCP for agent-led investigation, GitHub autofix for PR generation, or both from the same foundation.
Shared foundation: drop in the SDK
One import, one init call. Auralogs captures structured logs, errors, and unhandled exceptions so both workflows start from the same production context.
Path 1: MCP investigation
Give Codex, Claude, Cursor, or your internal agent read-only access to production logs through MCP. No GitHub connection, source access, or PR permissions required.
Path 2: Autofix PRs
Connect GitHub only when you want source-aware automation. Auralogs can analyze an incident, inspect the relevant repo, and open a pull request for review.
Use cases for logging-first and agent-native teams
Different teams enter from different pains: agent debugging, error investigation, or PR generation. Each path starts with the same structured logs.
Let your AI agent debug production logs
Connect Codex, Claude, Cursor, or an internal agent to Auralogs. It can search production logs, inspect metadata, and explain failures without copy-paste. Read keys are scoped to one project and do not grant source or write access.
Read use caseFind root cause faster
Auralogs gives teams searchable production logs, severity timelines, trace metadata, and optional AI summaries, so you can move from alert to evidence before opening another tool.
Read use caseTurn production incidents into reviewed fix PRs
Connect a specific GitHub repo and your own AI provider key only for source-aware automation. Auralogs maps incidents back to source code and opens pull requests for review, while logging and MCP keep working without GitHub.
Read use caseSDKs for the stacks you already ship
import { init, auralogs } from 'auralogs-sdk'
init({ apiKey: 'aura_...' })
auralogs.error('Payment failed', {
userId: 'usr_123',
error: 'card_declined'
})Payment processing failing for user usr_123
Stripe returned card_declined. This user has had 3 failed attempts in the last hour. Check if the card on file has expired. Consider notifying the user to update their payment method.
JavaScript
Browser, Node, Workers
npm install auralogs-sdkPython
FastAPI, Django, Flask
pip install auralogsJava
JVM, SLF4J bridge
ai.auralogs:auralogs-coreSwift
iOS, macOS, SwiftUI
Swift PackageGo
Services and CLIs
go get github.com/auralogs-ai/auralogs-goRust
tracing and panic capture
cargo add auralogsC++
C++17, libcurl transport
CMake packageGodot
Godot 4.5+ games
Godot addonPrefer agent-native setup? Copy a prompt for Codex or Claude, then add your Auralogs key after signup.
Simple, transparent pricing
Start free with structured logs and MCP access. Add AI analysis, alerts, and GitHub autofix as optional add-ons. Autofix uses your own Anthropic or OpenAI key.
Free
- 1 project
- 50,000 logs/month
- 7-day retention
- 1 team seat
- MCP server access
- Optional autofix PRs (BYOK)
- Email alerts
- Community support
Starter
- 3 projects
- 250,000 logs/month
- 30-day retention
- 3 team seats
- MCP server access
- Optional autofix PRs (BYOK)
- Email + webhooks
- Email support
Pro
- Unlimited projects
- 2,000,000 logs/month
- 90-day retention
- Unlimited team seats
- MCP server access
- Optional autofix PRs (BYOK)
- Email + webhooks
- Priority email support