Explorations
Strategic discussions, working theses, and things we're trialing — not yet settled positions. Newest first.
- Fri, Jun 5, 2026
Multi-Agent Autonomous Development: Runbook
Operational companion to the multi-agent autonomous development strategy doc. This is how you actually use the stack: one-time setup (both binaries on PATH, three GitHub credentials, sound hook), the per-task loop (one cmux pane kicks off aileron-agent task <slug>; brainstormer runs interactively, then the chain detaches; two commands in steady state), the soak period (during which every chain holds-for-review until you opt into auto-merge), parallel tasks across cmux workspaces, honest gotchas (foreground brainstorm + background chain, one container per worktree per phase, walks-away is a hypothesis), and what's still deferred (passive watcher daemon, librarian).
- Wed, Jun 3, 2026
Multi-Agent Autonomous Development: Aileron's Internal Build Stack
Aileron ships faster by running many specialized AI agents in parallel, each in its own microVM, each with a defined role — planner, coder, reviewer — connected through typed handoffs and a single blocking gate (tests). The thesis: Claude reasons and reviews, Codex writes, the VM is the trust boundary, cmux is the surface. The constraint from Cognition's 2025 post-mortem: writes stay single-threaded per worktree, full traces (not summaries) cross every handoff. We are customer-zero for our own v4 runtime — the shell-mediation, credential-sealing, action-contract substrate we sell is exactly the substrate this stack needs at scale.
- Mon, Jun 1, 2026
Describe it. Automate it. Done.
Write a repeatable workflow in plain English, built on your actual systems, runnable on day one. That's your Flight Plan. Share them and everyone flies faster.