Cairn
GitHubDashboard

Roadmap & help wanted

These are the bigger projects we'd love help with. Each one is a meaningful, self-contained contribution that improves the self-hosting experience. They're larger than a typical bug fix — good if you want to own something substantial.

Before you start: comment on (or open) the matching issue so we can align on approach and nobody duplicates work. See Contributing for the dev setup and PR flow. Browse the help wanted label for the live list.

One-command self-host

Bundle the whole stack behind a single docker-compose up — web + worker + Postgres (and the dev background-job runner) — so a newcomer can go from clone to running Cairn without provisioning anything by hand.

  • Why: today self-hosting means standing up several services separately; this is the single biggest barrier to trying Cairn.
  • Where to start: apps/web and apps/worker already ship Dockerfiles; this is mostly a docker-compose.yml plus wiring env defaults and a Postgres service. See Self-hosting.

Model-provider abstraction

Flow mapping and issue drafting are currently hardwired to Google Gemini. Introduce a small provider interface so self-hosters can choose Claude, OpenAI, or another model with a config switch.

  • Why: self-hosters shouldn't be forced onto one model vendor; an abstraction also makes the AI layer easier to test and extend.
  • Where to start: the AI calls live in the runner / issue-writer and the mapping routes under apps/web/app/api. Factor the provider behind an interface and select it via an env var.

Reduce mandatory managed services

A self-hoster currently needs Inngest (scheduling) and Cloudflare R2 (video / trace storage). Add self-hostable defaults — e.g. a local/cron scheduler and a filesystem or S3-compatible storage driver — so neither is required.

  • Why: fewer mandatory third-party accounts means a genuinely self-contained deployment.
  • Where to start: storage already has a driver concept (STORAGE_DRIVER=volume); scheduling needs an Inngest-free path.

Deploy recipes beyond Railway / Vercel

Document and verify deployments on more platforms — Fly.io, Render, a bare VPS, Kubernetes — with copy-pasteable config.

  • Why: people want to run Cairn where they already operate; the more first-class targets, the lower the adoption bar.
  • Where to start: generalize the existing Railway/Vercel notes in Self-hosting into per-platform guides.

Don't see your idea here? Open an issue — proposals for new flows, integrations, and DX improvements are all welcome.