create-syrinx-agent 4.4.0

package/LICENSE

@@ -0,0 +1,22 @@
+MIT License
+
+Copyright (c) 2026 Kuralle
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+

package/README.md

@@ -0,0 +1,91 @@
+# create-syrinx-agent
+
+Scaffold a [Syrinx](https://github.com/kuralle/syrinx) voice agent project.
+
+```bash
+npm create syrinx-agent my-agent -- --stt deepgram --reasoner aisdk --tts cartesia
+```
+
+One generator with conditional emission, not a directory of templates. You pick a provider
+per pipeline stage and get back a project that depends on **exactly those** providers, with
+a `.env.example` naming exactly the keys they need.
+
+## Pick a pipeline
+
+**Cascade** — speech in, text through a reasoner, speech out:
+
+```bash
+npm create syrinx-agent my-agent -- \
+  --stt deepgram --reasoner aisdk --tts cartesia
+```
+
+| Flag | Options |
+| --- | --- |
+| `--stt` | `deepgram` · `google` · `elevenlabs` · `grok` |
+| `--tts` | `cartesia` · `elevenlabs` · `gemini` · `openai-tts` · `grok` |
+| `--reasoner` | `aisdk` (any `ai@6` model) · `kuralle` · `mastra` |
+| `--vad` | `silero-vad` — optional; provider STT owns it by default |
+| `--endpointing` | `pipecat-smart-turn` · `vap` — optional |
+
+**Speech-to-speech** — one model hears and speaks, no separate stages:
+
+```bash
+npm create syrinx-agent my-agent -- --realtime realtime
+```
+
+`--realtime` is exclusive with `--stt`/`--tts`, and passing them together is refused rather
+than silently ignored — a speech-to-speech pipeline has no separate stages to configure.
+
+## Transport and runtime
+
+```bash
+--transport browser|twilio|telnyx|smartpbx     # default: browser
+--runtime   node|cloudflare                    # default: node
+```
+
+`--runtime cloudflare --transport telnyx` generates, but warns: Telnyx is not yet bound on
+the Workers edge, so that combination is deploy-unverified.
+
+## Presets
+
+Flag bundles, and an explicit flag always wins:
+
+```bash
+npm create syrinx-agent support-line -- --preset phone --tts elevenlabs
+```
+
+## Other options
+
+```
+--name <project-name>   default: the target directory's basename
+--yes                   accept defaults, never prompt
+--no-install            write the project, skip npm install
+--dry-run               print the file list and exit
+```
+
+## What you get
+
+```
+src/agent.ts            createAgent() — the --agent seam's factory export
+scripts/dev-server.ts   pnpm dev — local server, mic in the browser
+test/fixtures/          a fixture for the headless check
+.env.example            exactly the keys your combination needs
+AGENTS.md               what a coding agent can and cannot verify here
+package.json            only the providers you chose
+```
+
+The generated project wires to the same `--agent <module>#<export>` seam that Syrinx's dev
+server and [`@kuralle-syrinx/cli`](https://www.npmjs.com/package/@kuralle-syrinx/cli) use, so
+`pnpm dev` and `syrinx turn` drive the identical agent.
+
+## Two operators
+
+The generated `AGENTS.md` is written for a coding agent, and it is explicit about the line
+it cannot cross. An agent can assert transcripts, exit codes and latency numbers. It cannot
+judge whether barge-in feels natural, whether a pause reads as thinking or as a hang, or
+whether a voice sounds warm. Those need a person listening in the studio, and pretending
+otherwise is how a green check becomes false confidence.
+
+## License
+
+MIT