Overview

BreachLine runs Nebula, an AI penetration-testing engine, and exposes it through four surfaces. They share one account, one credit wallet and one set of permissions, so a key you create in the console works for the API and the CLI alike.

The surfaces

Surface	Use it to	Where
LLM API	Call Nebula models from your own code with OpenAI-compatible requests, or Anthropic-compatible ones (beta).	https://api.breachline.io/api/v1/llm/v1
Platform API	Manage API keys, read usage, set a spend cap, export your data and send integration notifications.	https://api.breachline.io/api/v1
CLI	Run autonomous, operator-confirmed testing from your terminal or CI.	breachline on macOS and Linux
Console	Run sessions, triage findings, build reports and administer your organization.	The BreachLine web app

Integrations connect the console and Nebula to Slack, Jira, source control, workspace suites, recon data providers and your own HTTP APIs.

Base URLs

All API traffic goes to one origin, https://api.breachline.io, over HTTPS.

• OpenAI SDKs take the base URL https://api.breachline.io/api/v1/llm/v1. The SDK appends paths such as /chat/completions.
• Platform endpoints live under https://api.breachline.io/api/v1/, for example /api/v1/keys/list.
• Anthropic SDKs take the base URL https://api.breachline.io/api/v1/llm for the Anthropic-compatible API (beta). The SDK appends /v1/messages.

Note

The LLM API also answers on /api/v1/llm/<path> without the /v1 segment, as an alias for SDKs that add their own version prefix. Prefer the full /api/v1/llm/v1 base in new integrations.

How requests are authorized

Every request carries a BreachLine API key (bl_live_ followed by random characters) or a console session token. A key’s scopes limit what it can do on the LLM API and with key management; on the other Platform endpoints, a key acts with the member scopes of the person who owns it. See Authentication and API keys.

How usage is billed

Your organization holds a prepaid credit wallet in GBP. LLM requests check for spendable credit before any model runs and are metered per token afterwards. See Billing and credits.

Where to go next

QuickstartCreate a key and make a first request.ModelsPick the right Nebula model.API referenceEvery public endpoint, by version.