Skip to content

Production debugging for AI agents

The agent-native debugging layer for modern products

DebugBundle captures production incidents, assembles deterministic debugging context, and delivers it through API, CLI, MCP, dashboards, and machine-readable artifacts so humans and AI agents can diagnose failures faster.

Quick installation guide

Choose the agent path, or install the CLI and integration manually.

Full installation guide

Agent Path

Set up DebugBundle for this repository end to end. Inspect the project language, framework, package manager, entry points, logging setup, and whether it has a frontend, browser surface, or server-rendered UI. Read the relevant DebugBundle docs: https://debugbundle.com/docs/installation, https://debugbundle.com/docs/quickstart, https://debugbundle.com/docs/project-setup/connect-to-cloud, https://debugbundle.com/docs/cli/cloud-workflow, https://debugbundle.com/docs/cli/local-workflow, https://debugbundle.com/docs/sdks, and https://debugbundle.com/docs/agent-workflows/skill-file. Run `debugbundle setup`, then read and follow `.agents/skills/debugbundle/SKILL.md`. Choose Cloud mode for hosted deployment or shared incident workflows; otherwise use local-only mode. If Cloud is appropriate, run `debugbundle login` and `debugbundle connect`. Install the smallest safe integration path: backend SDK, WordPress plugin, or log ingestion. For browser-capturable UI, also install the Browser SDK and configure same-origin relay for full-stack apps or direct capture for frontend-only apps. Configure required env vars or tokens. Trigger one intentional test error through the real ingestion path. Verify with `debugbundle verify cloud` or `debugbundle verify local`, then show the incident using `debugbundle incidents --source cloud|local` and inspect it with `debugbundle inspect <incident-id>`.

Use this prompt to let your coding agent set up and choose the right DebugBundle mode.

CLI

npm install -g @debugbundle/cli
debugbundle --version

The CLI runs setup, local processing, incident inspection, token management, cloud connection, and verification.

SDKs

npm install @debugbundle/sdk-node
import DebugBundle from '@debugbundle/sdk-node';

DebugBundle.init({
  projectToken: 'local',
  environment: 'production',
  service: 'api',
});
Framework exports include Express, Fastify, Next.js, and browser relay handlers. See Node.js SDK.

How it works

Step 1

Capture

SDKs capture exceptions, requests, logs, breadcrumbs, and probes from your app.

Step 2

Normalize

Events are redacted, fingerprinted, and grouped into incidents automatically.

Step 3

Bundle

DebugBundle assembles deterministic incident context with reproduction hints.

Step 4

Resolve

Humans and agents inspect the same bundle through dashboards, API, CLI, or MCP.

Install the path that matches your stack

Start with the CLI, then add the smallest SDK or ingestion path for your app.

Built for agents

  • Human and machine readable

    Structured JSON bundles, llms.txt, OpenAPI, JSON Schemas, and stable examples.

  • Agent-native interfaces

    API, CLI, and MCP expose the same workflows. No important capability is dashboard-only.

  • Same source of truth

    Docs and references are generated from the contracts that power the product.

Production ready

  • Local-first by default

    Capture, process, and inspect incidents locally before connecting hosted services.

  • Deterministic bundles

    The same events produce byte-identical output with no random IDs or generation-time clocks.

  • SDK safety guarantee

    SDKs do not throw uncaught exceptions, block requests, or crash the host process.

Every interface, same capabilities

DebugBundle enforces interface parity. If a capability matters for automation, it is available through API, CLI, and MCP — not locked behind a dashboard.

Ready to debug smarter?

Start locally, or use the hosted free tier when you want cloud ingestion. Upgrade when you need team collaboration, longer retention, and advanced features.