LogPulse
LogPulse
Get Started
For apps built with Claude Code, Cursor & Codex

Your agent shipped it. Now close the loop.

LogPulse watches your app in production for attacks, errors and slow endpoints, and gives what it finds back to the coding agent that wrote the code. The agent proposes the fix; you approve it.

Start freeSee how it works
  • Free plan, no credit card
  • Set up in about ten minutes
  • Nothing changes without your approval
  • EU-hosted, GDPR by default

Code ships faster than anyone reads it

A coding agent writes the API, the database schema and the deploy config in an afternoon. Then the app is live and nobody is watching: not the agent, which never sees production, and usually not you either.

~45%

of AI-generated code samples contained an OWASP Top 10 weakness

Source: Veracode, 2026
1 in 10

scanned Lovable apps left database tables open (CVE-2025-48757)

Source: public audit of 1,645 apps
2×

the rate of leaked secrets in commits co-written by a coding agent

Source: GitGuardian, 2026

Close the loop

Build, ship, watch, fix: one loop, one agent

The agent that wrote the code is the best one to fix it, as long as it can see what happens in production. LogPulse gives it that view, with you in the approval seat.

  1. 1

    Build

    Your coding agent writes the app, and with the LogPulse MCP server it also adds structured request logging.

  2. 2

    Ship

    Deploy as usual. Every request lands in LogPulse with the fields that health checks and security detections need.

  3. 3

    Watch

    Detections catch brute force, path scanning and injection attempts; service KPIs and anomaly detection catch slow or failing endpoints.

  4. 4

    Fix

    Ask your agent what happened. It reads production over MCP, fixes the code and proposes alerts or detections that you approve.

What LogPulse watches once you ship

Security

  • Brute force and password spraying on your login
  • Path scanning and probing of admin endpoints
  • SQL injection and path traversal attempts
  • One client pulling far more data than a user should

Performance and errors

  • Error rate and p95 latency per service, as KPIs
  • Anomalies against each KPI's normal daily pattern
  • Before-and-after comparison around every deploy
  • Uptime checks on your public endpoints

Set up in about ten minutes

The onboarding in LogPulse walks you through it and creates the keys for you.

  1. Create a free account

    No credit card. The tour creates an agent token and an ingest key.

  2. Connect your agent

    One command for Claude Code; the same URL and token for Cursor, Codex and any other MCP client.

  3. Let the agent build the logging

    Paste the prompt from the tour. The agent adds request logging with the fields detections and KPIs need, then proposes the detections and alerts.

Claude Code
claude mcp add logpulse --transport http \
  https://api.logpulse.io/mcp \
  --header "Authorization: Bearer lpat_…"
Then ask your agent

"Set up production logging with LogPulse for observability and security, and propose the detections a public web app needs."

Your agent proposes. You decide.

Propose and approve

Every change an agent makes is a proposal: checked against your real data first, applied only after a person approves it, and revertible.

Scoped and audited

Tokens carry only the scopes you give them. Every tool call is rate-limited and written to the audit log.

EU-hosted

Log data stays in the EU (GCP Amsterdam). Log content is treated as data, never as instructions.

Free for a new app

1 GB of logs a day, 7 days of retention, security monitoring, service health for 3 services, 5 alert rules and the MCP server. When the app grows, paid plans add volume and retention; nothing you set up changes.

Start freeSee pricing

Questions

What does "close the loop" mean?
Building and shipping are one half; knowing how the app behaves in production is the other. LogPulse collects your app's logs, runs security detections and health checks on them, and lets the coding agent that built the app read the results over MCP, so it can fix the code and propose alerts.
Which coding agents work with it?
Any MCP client: Claude Code, Cursor, Codex, VS Code with Copilot and others. They connect to one HTTPS endpoint with a token or OAuth sign-in.
Is it really free?
Yes. The Free plan has no time limit and needs no credit card. It covers 1 GB of logs per day with 7 days of retention, security monitoring, observability for up to 3 services, 5 alert rules and the MCP server.
Can the agent change my production setup?
Not on its own. Agents can only propose detections, alert rules, services and similar changes. LogPulse validates each proposal against your data and waits for a person to approve it, unless you explicitly allow a kind of change to apply automatically.
How do my logs get into LogPulse?
Your app sends them over HTTPS (plain JSON) or OpenTelemetry (OTLP). The onboarding gives your agent a prompt to add this to your backend or edge functions. A server or Kubernetes cluster can use the LogPulse agent instead.

Ship it, then let your agent watch it

Start free, connect your agent and see your first production logs today.

Start free