v0.4.0

What's Changed

Features

• Read intent out of a dictated answer, and give the gate a car by @jaypetez in #25
• End an answer on the trigger word, not on a pause by @jaypetez in #28
• Say the result out loud, and let the finish word be changed by @jaypetez in #30
• Add a mobile idea library and native sharing by @jaypet3 in #39
• Let the user choose the model, per provider by @jaypetez in #42

Fixes

• Say the wrap-up advisory once, not on every turn afterwards by @jaypetez in #26
• Script a recogniser, and settle a session that goes deaf by @jaypetez in #27

Voice and dictation

• Move the hands-free loop somewhere it can fail safely by @jaypetez in #29
• Prove the whole loop by voice, against the real app by @jaypetez in #31

Providers

• Drive a whole interview by voice against a real model by @jaypetez in #32

Documentation

• Correct the release doc: the package came up public without being asked by @jaypetez in #9
• Document driving mode where the next person will look for it by @jaypetez in #33
• Name both tripwires that cause an amber run, and the rate by @jaypetez in #34
• Support Claude Code and GitHub Copilot workflows by @jaypet3 in #35
• Add repo-local development skills and review agents by @jaypet3 in #38
• Publish comprehensive IdeaForge documentation by @jaypet3 in #40

Build and CI

• Harden architecture and delivery checks by @jaypet3 in #36
• Add shared release skill and prepare v0.3.1 by @jaypet3 in #37
• Prepare v0.4.0 by @jaypetez in #43

New Contributors

• @jaypet3 made their first contribution in #35

Full Changelog: v0.3.0...v0.4.0