Agentic engineering in your own repository
Claude Code is at its best when it works the way your team already works. This track takes you past autocomplete and into delegating real units of engineering — planning, implementing, reviewing and shipping — inside a codebase that has history, tests and opinions.
Ordered the way we teach it. Private workshops reorder and reweight this against what your team already knows.
Project memory, conventions files and context budgets — teaching the model your codebase once instead of re-explaining it every session.
The loop that survives a real repository: scope a change, let the agent draft it, then review with the same rigour you'd apply to a colleague.
Splitting a task across worktrees and specialised subagents, and knowing when the coordination cost outweighs the speed-up.
Making automation safe by construction: what the agent may touch, what needs a human, and how to enforce that in CI.
Wiring in your own tools through MCP servers and reusable skills, so the model can reach your services, not just your files.
Tracking diff quality, test outcomes and review time so you can prove the change in throughput rather than assert it.
Teams often pair a model track with Agentic Systems — the tool design and evaluation material applies whichever provider you settle on.
Send over your stack, your group size and roughly when you'd like it. We'll come back with a curriculum outline and a date.