Use this path for notes on working with AI coding agents on non-trivial tasks — analysis, refactors, and reviews — and the verification habits that make their output trustworthy.
Give an agent the scary work only behind a gate it cannot fake — tests, builds, reproducible checks.
Prefer many small verified steps over one large unverified change.
Keep an honest ledger of what was and was not covered.
A field note on handing a large Rust refactor to Fable: knowledge-graph analysis, a full graph rebuild that survived a mid-run failure, and a 9,859-line file cut to 877 across six verified commits.
A decision memo on moving from agent-by-agent coordination to Firstmate and Herdr: isolated worktrees, durable state, supervised loops, decision gates, and evidence-backed PR handoffs.
The Autonomy alpha is live — a native macOS app and local MCP server that lets AI agents work through your real accessibility setup — plus the day-three build-log of shipping it with Claude Fable 5.
