I build open-source platform engineering tools for small teams that work with AI coding agents, and I build them in public. Each stack is small enough to run on a laptop, does one job, and states what it doesn't do yet.
Every repo below is generated from uFawkesAI and develops inside its
devcontainer, ghcr.io/paruff/fawkes-space.
| Repo | What it is | Where it is |
|---|---|---|
| 🤖 uFawkesAI | Template for AI-native development: one config for Claude Code, OpenCode, Codex and Gemini CLI; an intent → spec → plan chain checked in CI; agent evals that gate merges; and the shared devcontainer image |
v2.0.0 in release candidate |
| 📊 uFawkesObs | Docker Compose observability stack: OpenTelemetry, Prometheus, Loki, Tempo, Grafana. Computes delivery (DORA) metrics from pipeline events | v0.2.0, heading to v1.0.0 |
| 🔧 uFawkesPipe | Woodpecker CI pipeline contract with security scanning, for any language | v1.7 beta, heading to v2.0.0 |
| 🛠️ uFawkesDevX | Developer portal and golden-path templates | Pre-release, heading to v0.1.0 |
| 🥋 uFawkesDojo | Belt-level platform engineering curriculum; each lab is built on a released stack | 0.1.0-alpha |
| 🦅 fawkes | Kubernetes internal delivery platform (Tekton, Argo CD, Grafana) for teams that outgrow Compose | Pre-alpha, heading to Tracer Bullet Alpha |
ufawkes.dev has a page for each stack, the compatibility matrix of which versions were tested together, and a live release status.
Releases are gate-driven, not date-driven: a stack ships when its acceptance criteria pass, and a claim that can't be verified gets removed instead of rushed. AI agents implement; I decide and review.
Based in Óbidos, Portugal 🇵🇹




