Trials, not promises.

The original experiments: what worked, what failed, and what it cost. These pages render directly from docs/trials in the repository. Historical records retain the name Tenon.

All trials, newest first
Trial	Source
Trial 0013 — The notes app again, with --with auth (ADR 0030)	Repository record
Trial 0012 — Correctness where agents often go wrong (the notes app)	Repository record
Trial 0011 — 0.3 with --select / --forms and the scaffold's text list (ADR 0029)	Repository record
Trial 0010 — 0.3 (map, composable scaffold, recipes), two runs per arm	Repository record
Trial 0009 — 0.2, measured with Claude Code in the app directory, Nuxt re-run the same way	Repository record
Trial 0008 — Build and change an app created from the published 0.1.0	Repository record
Trial 0007 — Build and change after ADR 0022 (a smaller authoring surface)	Repository record
Trial 0006 — Build and change after Tiers 2–4 (Phases 7–9)	Repository record
Trial 0005 — The change path after ADR 0013 iteration 2	Repository record
Trial 0004 — Re-running trial 0003 with the Tenon skill (ADR 0013)	Repository record
Trial 0003 — Fresh agents build and change the same app in Tenon and in Nuxt	Repository record
Trial 0002 — Is Tenon actually easier for an AI to write and maintain? (before Phase 6)	Repository record
Trial 0001 — writing a small blog with Tenon (AI authoring + SEO)	Repository record