Roadmap#

This document outlines the development roadmap for at-rund.

Current Status: Alpha (Under Heavy Construction)#

The core architecture is in place. Dev mode works on macOS/Linux with Nix. Production isolation is being implemented with multiple backends.

See DESIGN.md for architectural decisions.

Phase 1: Core Functionality ✅#

Goal: A working end-to-end system where bundles can be fetched from a PDS and executed.

ATProto Integration#

• DID resolution (did:plc, did:web)
• PDS client for fetching bundle records
• Bundle blob fetching and caching
• Manifest parsing (permissions, runtime, limits)

Bundle Execution#

• Wire up executor to HTTP routes
• Permission enforcement (net, read, write, env)
• Resource limits (memory, CPU, timeout)
• Secrets decryption and injection

Dev Mode#

• Nix-based execution (NixPool)
• Auto-detection of capabilities
• Runtime executor pattern (at-run-exec)
• Hot reload for local development
• Better error messages

Phase 2: Production Isolation#

Goal: Multiple isolation backends to balance security vs. accessibility.

isolation = "auto" | "none" | "container" | "firecracker"

Container Backend (In Progress)#

• ContainerPool executor implementation
• OCI image building via Nix (debian-slim base)
• Docker/Podman runtime detection
• seccomp profiles for syscall filtering
• Network namespace isolation
• Permission enforcement via container config

Firecracker Backend (Future)#

• FirecrackerPool executor implementation
• Kernel + rootfs image building via Nix
• VM lifecycle management (spawn, stop, reuse)
• virtio-fs for bundle mounting
• vsock for host ↔ guest communication
• Guest agent (Go binary inside VMs)

Shared Infrastructure#

• Auto-detection logic (KVM → container → none)
• Pre-warming (configurable per runtime)
• Idle timeout and reclamation
• Max instance limits
• Graceful drain on shutdown
• Network proxy with permission enforcement

Phase 3: Observability#

Goal: Operators can monitor their runners effectively.

Metrics#

• OpenTelemetry integration
• Request count, latency, error rate
• Per-bundle, per-DID breakdowns
• VM pool utilization
• Resource usage (memory, CPU)

Logging#

• Structured JSON logs
• Request tracing (trace IDs)
• Bundle execution logs (opt-in)

Dashboard#

• Example Grafana dashboard
• Prometheus scrape endpoint (/metrics)

Phase 4: Operator Experience#

Goal: Make it easy to run a production at-rund instance.

Deployment#

• systemd service support
• Docker image
• Nix flake for NixOS deployment
• Ansible/Terraform examples

Configuration#

• Config validation on startup
• Reload config without restart (SIGHUP)
• Environment variable overrides

Security#

• Security hardening guide
• Firewall recommendations
• TLS termination examples (nginx, caddy)

Phase 5: Advanced Features#

Goal: Features for larger-scale or specialized deployments.

Tasks & Jobs#

• Port task queue from at-run v1
• Background job execution
• Cron scheduling
• Result caching

Multi-Node#

• Shared state (Redis, SQLite)
• Load balancing considerations
• Sticky sessions for stateful bundles

Custom Runtimes#

• Runtime marketplace/registry (community-contributed)
• Documentation for writing runtimes
• Testing framework for runtimes

Phase 6: Ecosystem#

Goal: at-rund becomes part of a thriving ecosystem.

Discovery#

• Runner announcement protocol (optional)
• Capability advertisement (runtimes, limits)
• Uptime/health signaling

Developer Experience#

• at-run test --runner <url> for testing against remote runners
• Bundle compatibility checker
• Performance profiling

Documentation#

• Operator guide
• Security model explanation
• Troubleshooting guide
• Video tutorials

Non-Goals (For Now)#

These are explicitly out of scope for the initial releases:

• Automatic runner discovery — Trust is social; discovery is manual
• Payment/billing integration — Use middleware if needed
• Multi-region orchestration — Each runner is independent
• Bundle validation/signing — Trust the author, not the code
• Centralized registry — Bundles live on user PDSes

Contributing#

We welcome contributions! Areas where help is especially appreciated:

1. Runtime definitions — Create Nix configs for new runtimes
2. Testing — Run at-rund and report issues
3. Documentation — Improve guides and examples
4. Firecracker expertise — Help with VM integration

See CONTRIBUTING.md for guidelines.

Version History#