Demos

Demo Reels

Scripted terminal recordings of HELIX in action. Each demo runs in a Docker container for reproducibility and uses ddx agent run as the agent harness.

Quickstart: Full Lifecycle

The complete HELIX onboarding experience: install, frame, design, build, and review. Builds a Node.js temperature converter from scratch, driven entirely by HELIX artifacts and the tracker.

Act	Phase	What Happens
1	Install	Install HELIX skills and CLI, verify ddx agent harness
2	Setup	Initialize git repo, tracker, and AGENTS.md
3	Frame	Agent creates product vision, PRD, and feature spec
4	Design	Agent creates technical design, then tracker issues
5	Build	Red: write failing tests. Green: implement to pass.
6	Verify	Run tests, check acceptance criteria
7	Review	Agent reviews all work for gaps

docker build -t helix-demo docs/demos/helix-quickstart/
docker run --rm \
  -v ~/.claude.json:/root/.claude.json:ro \
  -v ~/.claude:/root/.claude \
  -v $(pwd):/helix:ro \
  -v $(pwd)/../ddx/ddx:/usr/local/bin/ddx:ro \
  -v $(pwd)/docs/demos/helix-quickstart/recordings:/recordings \
  helix-demo

Concerns: Preventing Technology Drift

Demonstrates how concerns keep agents on the declared technology stack. A Bun/TypeScript project declares typescript-bun as its concern. The agent builds with Bun-native tools, then a deliberate drift (vitest import) is introduced and caught by concern-aware review.

Act	Phase	What Happens
1	Setup	Create Bun project with `typescript-bun` concern
2	Frame	Agent reads concerns, uses Bun-native requirements
3	Build	Agent implements with Bun.serve(), bun:test, Biome
4	Drift	Deliberate Node.js drift introduced (vitest import)
5	Review	Agent detects drift and files a tracker issue

docker build -t helix-concerns-demo docs/demos/helix-concerns/
docker run --rm \
  -v ~/.claude.json:/root/.claude.json:ro \
  -v ~/.claude:/root/.claude \
  -v $(pwd):/helix:ro \
  -v $(pwd)/../ddx/ddx:/usr/local/bin/ddx:ro \
  -v $(pwd)/docs/demos/helix-concerns/recordings:/recordings \
  helix-concerns-demo

Evolve: Threading Requirements Through the Stack

Demonstrates how helix evolve propagates a new requirement through every layer of the artifact stack. Starting from a working temperature converter, a “Add Kelvin support” requirement is threaded through the PRD, feature spec, and technical design, then implemented via TDD.

Act	Phase	What Happens
1	Setup	Working v1 project with existing artifacts and code
2	Evolve	Thread “Add Kelvin” through PRD, feature spec, design, tracker
3	Build	TDD: failing Kelvin tests (Red), then implementation (Green)
4	Verify	All tests pass, new CLI flags work

docker build -t helix-evolve-demo docs/demos/helix-evolve/
docker run --rm \
  -v ~/.claude.json:/root/.claude.json:ro \
  -v ~/.claude:/root/.claude \
  -v $(pwd):/helix:ro \
  -v $(pwd)/../ddx/ddx:/usr/local/bin/ddx:ro \
  -v $(pwd)/docs/demos/helix-evolve/recordings:/recordings \
  helix-evolve-demo

Experiment: Metric-Driven Optimization

Demonstrates how helix experiment optimizes code through measured iteration. A deliberately slow string processing function is improved through hypothesis-driven changes, with correctness tests enforced at every step.

Act	Phase	What Happens
1	Setup	Project with slow function, correctness tests, and benchmark
2	Iterate	Agent hypothesizes, edits, tests, benchmarks (iteration 1)
3	Iterate	Agent finds next bottleneck, optimizes further (iteration 2)
4	Results	Final benchmark shows improvement, all tests still passing

docker build -t helix-experiment-demo docs/demos/helix-experiment/
docker run --rm \
  -v ~/.claude.json:/root/.claude.json:ro \
  -v ~/.claude:/root/.claude \
  -v $(pwd):/helix:ro \
  -v $(pwd)/../ddx/ddx:/usr/local/bin/ddx:ro \
  -v $(pwd)/docs/demos/helix-experiment/recordings:/recordings \
  helix-experiment-demo

Glossary