Engineering Onboarding

Your first week as an engineer. Do the company onboarding first, then this.

Set up

• macOS recommended (team is mostly Darwin/aarch64).
• Install: uv (Python), bun (TypeScript), docker, gh, tailscale.
• Get on Tailscale and into 1Password (Honcho Engineering vault) from a founder.

Self-host the core repos

Clone each, follow its README/CLAUDE.md/AGENT.md, get it running locally. Log every snag in the friction log.

Repo	Goal
Honcho	Run API + deriver locally; hit /v3/peers/{id}/chat
Groudon	Run the control plane; provision one tenant
Kyogre	Our reference Honcho integration (Cloudflare)
Excadrill	Run one small benchmark end-to-end
Bloom	Consumer app on your local Honcho

Then read

• Honcho ADR — why Honcho is shaped this way (esp. the Peer Paradigm).
• Infra — hardware, cloud, networking, telemetry.
• Engineering System — the tools we use and how.
• Honcho specs — what’s being designed now.

Conventions

• Python: uv only, never pip. TypeScript: bun only.
• Specs first: non-trivial work gets a spec before code (spec → Ultraplan → phased PRs). See wiki conventions.
• Dry-run first: confirm dry-run output before running ops scripts.
• --prod is a checkpoint: every Groudon script takes --test or --prod.
• Never commit certificate/, credentials/, or .env files.

Week-1 success

• Self-hosted Honcho and ran a benchmark.
• Shipped (or drafted) one fix from the friction log.
• Can explain the Peer Paradigm to someone new.