B
Offline
Back to Tutorial Paths
Highest-Leverage Path

Agent-Assisted Research

Let your AI coding agent internalize and apply the Brenner methodology. The most powerful path for users with Claude Code or Codex.

~50 min•8 steps•Requires Claude Max or GPT Pro

Prerequisites

  • Claude Code CLI or Codex CLI installed
  • Active Claude Max or GPT Pro subscription
  • Familiarity with terminal commands

Tutorial Steps

Eight steps to agent-augmented research

1

Why Agent-Assisted?

~3 min

Understand the leverage of having AI internalize methodology, not just follow instructions.

Why agents that internalize methodology outperform prompt-following
The human-AI collaboration model for research
2

Prerequisites

~5 min

Verify you have Claude Code or Codex installed and ready to go.

How to verify your coding agent is installed
The subscription requirements for each tool
3

Clone into Agent Context

~5 min

Clone the BrennerBot repository and give your agent access to the codebase.

How to give your agent access to the Brenner corpus
The key files your agent will read
4

Agent Studies the System

~10 min

Have your agent systematically study the Brenner documentation and internalize the methodology.

How to have your agent build a working understanding
What the agent should extract from each document
5

Define Your Research Problem

~5 min

Formulate your research question and have the agent help refine it using Brenner criteria.

Brenner's criteria for good research questions
How to sharpen vague questions into testable ones
6

Agent Builds the Inputs

~10 min

Generate hypothesis slate, assumption ledger, and third alternatives using Brenner operators.

What artifacts the Brenner loop requires
How the agent uses operators to generate alternatives
7

Agent Runs the Brenner Loop

~10 min

Full loop execution: generate discriminative tests, rank by potency, identify exclusions.

How to design tests that distinguish hypotheses
What makes a test 'potent' vs. weak
8

Human Review

~5 min

Work through an interactive checklist to evaluate output quality before using results.

How to evaluate agent-generated research artifacts
Common failure modes and how to fix them

Ready to Begin?

Start with Step 1 to understand what makes agent-assisted research different, then proceed through the tutorial at your own pace.

Start TutorialTry Quick Start Instead