2.works
2.works / Projects

Things I’ve made.

Software, sound, stories and experiments. A selection of things made to be used, played and explored.

insideLLMs terminal example comparing baseline and candidate runs and identifying two regressions in dummy-model outputs.
LLM evaluation tooling02

insideLLMs

A Python library and CLI for comparing model behaviour through shared probes, recorded runs and inspectable differences.

Open sourceExplore project
GitHub card for dr-gareth-roberts/harness-engineering: “Open-source toolbox for harness engineering: composable primitives for building robust LLM/agent harnesses.”
Agent infrastructure03

Harness engineering

Composable building blocks for the software around an AI agent: context, tools, hooks, memory and telemetry.

Open sourceExplore project
Context Engineering Toolkit banner: a stacked context window above a dashed budget line, with the tagline “Pack, debug, red-team, and orchestrate LLM context windows”, 17 packages, and TypeScript plus Python.
Context window tooling04

Context engineering

A toolkit for deciding what goes into an LLM’s context window: pack, score, cache, debug and red-team it.

Open sourceExplore project
A red car drifting through a glowing ring on a space circuit at 480 km/h, with a cracked blue planet ahead and the HUD showing lap 1 of 3 in seventh place.
Space circuit racer06

Spaceracer

Two weekends, GPT-6 Astra, and a game so far above my coding ability.

Playable in the browserExplore project
A figure in a dark suit, helmet and gas mask sits at an orange laptop behind a heap of banknotes and drink cans, lit by a single work lamp, with small orange robots half-buried in the paper around him.
Single and music video07

Lambda Dragonfire

A four-minute single about Mythos: too dangerous for you, safe enough for Chase, and available on the enterprise tier.

Streaming on SpotifyExplore project