DebugBundle

llms.txt

How AI agents discover DebugBundle capabilities through the llms.txt standard.

DebugBundle publishes llms.txt at the site root for crawler and agent discovery. It is not a setup path for humans; it is a stable index that points LLMs to the current docs, generated references, schemas, and example bundles.

What is llms.txt?

The llms.txt specification is a convention for providing machine-readable context to AI systems. Similar to how robots.txt guides web crawlers, llms.txt gives agents a compact map of where to read next.

The file is a structured markdown document that tells AI agents:

  • What the product does
  • Where to find documentation
  • What APIs and tools are available
  • How to integrate

DebugBundle's llms.txt

DebugBundle's file is available at:

https://debugbundle.com/llms.txt

It includes:

Product Summary

A concise description of DebugBundle's purpose: capture runtime failures, group them into incidents, and publish deterministic debug bundles for humans and agents.

Structured links to current documentation:

  • Start here: quickstart, installation, how it works, core concepts
  • Setup paths: local-only, cloud connection, project profile
  • Capture paths: SDKs, browser relay, Python, PHP, Java, Go, Ruby, Android, iOS, React Native, WordPress, log ingestion
  • Interfaces: CLI, API, MCP, local and cloud workflows
  • Operations: bundles, incidents, probes, webhooks, alerts, GitHub automation
  • Generated references: API endpoints, CLI commands, MCP tools, schemas, errors

Integration Points

Stable machine-readable links:

  • openapi.json
  • schemas/bundle.json
  • schemas/profile.json
  • schemas/webhook-events.json
  • schemas/mcp-tools.json
  • Example failure and improvement bundles

Agent Guidance

The file tells agents to prefer MCP when available, use the CLI for local-only projects or repositories with a .debugbundle/ directory, use the API for custom automation, and use the browser relay path when adding browser capture to projects with a backend.

How Agents Use It

Discovery

When an AI agent encounters a DebugBundle-instrumented project, it can fetch llms.txt to understand:

  1. Which setup path applies: local-only or connected cloud
  2. Which capture path applies for the project language or framework
  3. Whether CLI, API, or MCP is the right interface
  4. Which generated reference or schema to read for exact contract details

Example Agent Flow

1. Agent detects `.debugbundle/` or a DebugBundle package in a project
2. Agent fetches https://debugbundle.com/llms.txt
3. Agent follows the local-only, connected cloud, SDK, or relay docs for the stack
4. Agent uses MCP tools or `debugbundle inspect <incident-id>` to investigate issues

MCP Integration

Agents with MCP access usually do not need to read llms.txt after discovery. The MCP tool descriptions provide task-specific structured metadata. llms.txt is most useful when an agent starts from the public web or needs a compact list of stable DebugBundle URLs.

Generation

The root file is generated by the core public-artifact pipeline and vendored into the static site. Update the generator when public docs routes, schemas, or agent flows change.

Next Steps

On this page

What is llms.txt?DebugBundle's llms.txtProduct SummaryDocumentation LinksIntegration PointsAgent GuidanceHow Agents Use ItDiscoveryExample Agent FlowMCP IntegrationGenerationNext Steps