How It Works
auralogs connects your application logs to the AI tools you already use. The two main workflows are independent: use MCP when you want an agent to investigate logs, and use autofix when you want GitHub-connected PR generation. You can enable either one without enabling the other.
Shared foundation
Section titled “Shared foundation”1. Your app sends logs
Section titled “1. Your app sends logs”The auralogs SDK captures logs, errors, and unhandled exceptions from your application. Logs are batched and sent to the ingest endpoint. Errors are sent immediately.
2. Ingest receives and stores
Section titled “2. Ingest receives and stores”The ingest worker at ingest.auralog.ai validates your ingest key, checks your plan quota, and stores logs in the database.
From there, choose the path that fits the job.
Path 1: MCP investigation
Section titled “Path 1: MCP investigation”Use this path when you want Codex, Claude, Cursor, or another agent to answer questions from production logs. It does not require GitHub access or autofix permissions.
1. Get a read key
Section titled “1. Get a read key”Copy the MCP/API read key from the new-project startup modal, or create another from Settings → API & MCP keys. Read keys are scoped to one project and can only read data; they cannot write logs, change settings, or access another project.
2. Connect your MCP client
Section titled “2. Connect your MCP client”Connect Codex, Claude, Cursor, or another MCP client to https://mcp.auralog.ai/mcp. The client can call tools to list logs, search messages, fetch full log payloads, and read analyses during a conversation.
You can also inspect the same logs in the dashboard log viewer without configuring an AI provider.
Use MCP when you want to ask questions like:
- “What changed around the first checkout error?”
- “Find recent fatal logs in production and group them by cause.”
- “Pull the last 20 errors for this trace ID and summarize the user impact.”
Path 2: Autofix PRs
Section titled “Path 2: Autofix PRs”Use this path when you want auralogs to connect log context to source code and open a pull request. It requires an AI provider key and a connected GitHub repo, but it does not require MCP.
1. Enable AI analysis
Section titled “1. Enable AI analysis”Add an Anthropic or OpenAI key under Settings → AI providers. When error or fatal logs arrive, auralogs can generate analyses with severity, root cause, and recommended actions.
2. Connect GitHub
Section titled “2. Connect GitHub”Connect a repository under Settings → Repository. auralogs only gets access to the repos you explicitly select.
3. Choose autofix mode
Section titled “3. Choose autofix mode”Set Auto-Fix Mode to Auto-Fix when you want auralogs to open PRs for actionable analyses. Autofix PRs are opened for review and are never auto-merged.
Optional outputs
Section titled “Optional outputs”Depending on your setup:
- Analyses — generated analyses appear in your project’s Analyses tab with severity, root cause, and recommendations.
- Email — you receive a notification with the analysis summary.
- Webhooks — send analysis data to Slack, Discord, or your own incident management tools.
- Autofix PR — if a GitHub repo is connected and autofix is enabled, your provider can open a pull request with the fix.
Scheduled reviews are part of the analysis/autofix path. In addition to real-time error analysis, your provider can periodically review recent logs to detect patterns, recurring issues, and trends that individual error analysis might miss.