Jan Brennenstuhl Principal Engineer, Berlin

Security that makes the secure path the easy one.

Essays, a field guide and research notes on application security, identity, and the agents that now act on our behalf.

Featured essay

What I Learned Building an Identity Broker With Coding Agents

Magnus Jungsbluth and I built the open-source Agentic Identity Broker with coding agents from the first prototype. Over 10 months, I switched agent harnesses 6 times and used 9 models, and most of my lessons are about specs, tests, and review.

10 min read Read the essay

Now

Recent essays

All writing RSS

  1. Agentic Engineering

    Agent Substrate Knows Which Agent Is Calling, but Not for Whom

    Agent Substrate 1.0, kagent and agentgateway form a Kubernetes agent platform that keeps credentials out of the sandbox, but every identity in it names only the agent. A proof of concept shows how the Agentic Identity Broker can add the person behind the agent, and that neither the platform nor the broker is ready for it yet.

    17 min read

  1. State of the Broker 7 min State of the Broker: October 2026
  2. Identity & Auth 9 min Measuring How MCP Servers Authorize Agents
  3. Agentic Engineering 6 min A Read-Only MCP Server with Bunny Edge Script
  4. AppSec for Agentic Engineering 5 min Guest Lecture at HTW Berlin: AppSec in the Age of Agentic Engineering
  5. AppSec for Agentic Engineering 6 min Shift Down: Why Agentic Engineering Demands Platform-Level Security

Field guide

All entries Glossary

  1. Agent identity

    Should an agent use a robot account or delegated access?

    When an agent acts for a person, give it delegated access that names both the person and the agent, with narrow scopes, an expiry, and a way to revoke it. A robot account …

    Not yet reviewed

  2. Account security

    What should a password policy require?

    Under NIST SP 800-63B-4, a password used on its own must have at least 15 characters, or 8 when it is one factor of multi-factor authentication, and systems should accept …

    Not yet reviewed

  3. OpenID Connect

    What is inside a JWT?

    A signed JWT is 3 base64url-encoded parts joined by dots: a header that names the algorithm, a payload of claims, and a signature over both. The payload is encoded, not …

    Not yet reviewed

  4. Agent identity

    How does OAuth 2.0 Token Exchange keep provider tokens away from agents?

    OAuth 2.0 Token Exchange (RFC 8693) lets a client trade one token for another token meant for a specific target service. When a gateway runs that exchange on each tool …

    Not yet reviewed

  5. Agent identity

    How should people grant and revoke an agent's access?

    Let people grant each agent access per service, with a permission set they choose and an expiry, and let them revoke that grant at any time. Revoking the grant has to …

    Not yet reviewed

  6. Agent identity

    How does authorization work for MCP servers?

    In the MCP authorization specification (revision 2026-07-28), a protected MCP server over HTTP is an OAuth 2.1 resource server: it publishes Protected Resource Metadata …

    Not yet reviewed

Projects

Open-source work at the intersection of security and engineering. All projects

  1. Open source since

    Letsauth Lab

    An open lab at letsauth.eu that measures how public MCP servers implement the MCP Authorization specification, for people who build or integrate agent authorization.

  2. Open source since

    Agentic Identity Broker

    Scoped, revocable access to tools for agents acting on behalf of people.

Topics

Essays, field-guide entries, talks and projects by subject. All topics Search

For agents

Read this site from your agent

A read-only MCP server lets Claude, ChatGPT or any other MCP client search the essays and the field guide, read entries as Markdown and quote findings with their sources. It needs no sign-in.

https://mcp.janbrennenstuhl.eu/mcp

search_site get_entry get_findings list_topics

How to connect