Skip to content

Lead the README with lk agent init, and let the agent detect the spoken language - #111

Open
bcherry wants to merge 19 commits into
mainfrom
bcherry/inspiring-ramanujan-n56lsd
Open

bcherry wants to merge 19 commits into
mainfrom
bcherry/inspiring-ramanujan-n56lsd

Conversation

@bcherry

@bcherry bcherry commented Oct 15, 2026 •

Copy link
Copy Markdown
Contributor

Why

The starter no longer bundles LiveKit's agent skills. A project created with GitHub's Use this template button or git clone ends up with no skills, and the README didn't make lk agent init the obvious way to start.

Trying the Voice AI quickstart with a coding agent also turned up places where AGENTS.md and the agent's prompt steer it wrong.

Changes

  • Getting started now comes right after the title. It's four numbered steps that end with you talking to your agent:

    1. Install uv.
    2. Install the LiveKit CLI and sign in to LiveKit Cloud, signing up first if needed (lk cloud auth).
    3. lk agent init my-agent --template agent-starter-python.
    4. cd my-agent, lk agent console. uv sync is only needed if lk agent init didn't install dependencies.
  • Removed the manual setup (git clone) and the CLI version and update notes.

  • The feature list moves under a What's included heading.

  • Using coding agents moves further down, after Testing and debugging. It has two items:

    • Agent skills: lk skills install.
    • AGENTS.md.

    The docs search item is gone.

  • Removed the Self-hosted LiveKit section. Deploying to production only covers LiveKit Cloud.

  • AGENTS.md: if the skills are missing, the coding agent asks the developer for permission, then runs lk skills install --yes. Until the next session picks them up, it reads the SKILL.md files directly.

  • AGENTS.md: the debugger's minimum CLI version is now 2.19.0.

  • AGENTS.md: simulations run with --yes, which a coding agent needs to add ids to new scenarios without a terminal.

  • AGENTS.md: when the agent's purpose changes, update the starter scenarios, which still expect the general-purpose assistant.

  • AGENTS.md: a passing simulation only covers its scenarios' inputs, so try other inputs in the debugger.

  • Agent prompt: the "spell out numbers" rule applies to responses meant for TTS, not tool arguments. It had made an agent save an address in words.

  • STT no longer sets a language. Universal-3.6 Pro detects and switches languages by itself: in an audio test, English transcribed the same, and Spanish came through intact only without en.

Companion changes: livekit-examples/agent-starter-node#72 (same changes), livekit/web#6598 and livekit/web#7044 (docs), and livekit/agent-skills#20.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu

The skills are no longer bundled, so a project created with GitHub's
"Use this template" button or git clone has none. Make lk agent init the
recommended way to create a project, add lk skills install to the manual
setup, lead the coding agents section with the skills, and tell coding
agents to recommend lk skills install when the skills are missing. Raise
the stated CLI minimum to 2.19.0, which lk skills and the debugger need.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
@bcherry bcherry changed the title Recommend lk agent init and explain how to get the agent skills Lead the README with lk agent init and make agent skills setup direct Oct 16, 2026
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rquu1YZenr2ZXTsyVptCYu
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NyEkbREWLMphp6SWj1Pk1u
@bcherry
bcherry marked this pull request as ready for review November 2, 2026 21:55
…arguments

- Simulations from coding agents need --yes to add scenario ids
- Update the starter scenarios when the agent's purpose changes
- A passing simulation only covers its scenarios' inputs; try others in the debugger
- The spell-out rule applies to speech only, so tool arguments keep their written form
- Note removing the STT language to detect and switch languages
- README: dependency install only if lk agent init skipped it
@bcherry bcherry changed the title Lead the README with lk agent init and make agent skills setup direct Lead the README with lk agent init, and let the agent detect the spoken language Nov 6, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants