Skip to main content
Skip to content
Menu

Codex Intermediate 3 parts Updated

Build a coding-agent capability matcher

You'll have a working classifier (script or spreadsheet formula) that routes your coding tasks to the right AI model based on complexity level, reasoning needs, and your team's benchmark data.

What you'll make

Coding Agent Capability Matcher

  1. assess_task_complexity() — classifier for task attributes: what to include
  2. route_to_tool() — decision logic with override rules: what to include
  3. benchmark_tracker() — performance logging and analysis: what to include
  4. main() — orchestration and usage interface: what to include

Upload your working Python script with example usage on real tasks

The 3 parts

Each part ends with a check you can see.

  1. Benchmark Your Current Stack

    Test 3+ AI coding tools on the same 5 tasks from your actual work to find performance patterns

    Done when

    You have performance data on 3+ tools across 5 real coding tasks, with specific notes on what each tool got right vs wrong

  2. Map Task Complexity Indicators

    Identify the specific task attributes that predict which tool will perform best

    Done when

    You have 3-5 measurable task attributes that reliably predict tool performance, with thresholds for each

  3. Build the Routing Decision Engine

    Code a working classifier that takes task attributes and outputs which tool to use with confidence levels

    Done when

    You have a working function that accepts task parameters and returns a tool recommendation with reasoning

    Capstone: you put together your Coding Agent Capability Matcher.

Where this helps

An elective: a single lesson that sits outside rtwo's job certificates and doesn't count toward one.

About this lesson

What you'll be able to do

  • Run a working classifier that routes your coding tasks to the right AI model based on complexity and reasoning needs
  • Have benchmark data on 3+ AI coding tools from your actual tasks, not just demos
  • Know exactly when to use which tool and when to switch mid-task
  • Have a repeatable testing method for any new AI coding capability that appears

Who it's for

  • Developers who use multiple AI coding tools but pick them randomly or by habit
  • Teams trying to standardize on AI tools without clear performance criteria
  • Anyone who's wondered if they're using the wrong AI for their coding tasks

What it won't cover

  • Setting up API access or authentication for specific tools
  • Prompt engineering techniques for individual models
  • Code review processes or team adoption strategies
  • Cost optimization across different AI pricing models

Skills you'll practise

  • Systematic capability benchmarking
  • Task complexity assessment
  • Decision rule codification
  • Performance pattern recognition