THE CODE PLANT · Observable software and AI agents
An ongoing personal R&D project: turn observed Python execution into a navigable representation and evaluate changes proposed by AI agents.
Project Description
I started turning this personal idea into a working implementation in October 2026. THE CODE PLANT is still under development and evolving; it is not a finished product.
The current work explores Python execution instrumentation, linking activity, calls, results and errors to source code, with graphs and 3D visualisation.
I am also developing isolated change comparisons, replay of affected paths and MCP tools for Claude/Codex agents. The goal is to support development decisions with execution evidence.
I am evaluating the approach through exercised execution paths and comparisons between versions to guide its next iterations.
Main Features
- Personal R&D project started in October 2026
- Python execution instrumentation
- Graphs and 3D visualisation
- Isolated change comparisons
- Replay and MCP tooling
- Scope and feasibility under evaluation