tough-tongue/toughtongue-skills / skills/ttai-agent
ttai-agent
VerifiedCreates, edits, and refines Tough Tongue AI Scenarios, the definitions behind voice agents (sales calls, lead qualification, demos, screening, support, booking) and practice roleplays (negotiations, interviews, coaching). Writes ai_instructions, ai_model_config, strategy, tools_config, and rubrik through ttai:create_scenario and ttai:update_scenario. Also resolves the personal or organization workspace, checks plan and balance, places or cancels SIP calls, schedules or cancels meeting bots, mints private Scenario links, grounds a Scenario in the user's documents with knowledge FileBases, and troubleshoots ttai MCP sign-in. Use when the user wants a voice agent or practice Scenario created or improved ("create a voice agent", "the agent ends calls too early", "make it sound more natural"), a call placed, a meeting bot sent, or a Tough Tongue AI account question answered. Not for reports across many sessions (ttai-session-analyst) or browser demo steps (ttai-browser-demo-builder).
references/scenario/model-selection.md
View on GitHub# Scenario model selection
How to stamp `ai_model_config`. Load the `ttai` tool schema before calling — it
is the type source of truth.
Sibling files: [control.md](control.md) (strategy, tools, appearance),
[ai-instructions.md](ai-instructions.md). On update, `ai_model_config` is
replaced whole — resend the full stamp, including `tts_voice_id`.
## Contents
- Three pipelines (realtime, separate TTS, Cascade)
- Need-based overrides
- Situation → starting stamp
- Cartesia voice IDs
- Catalog families
---
Always set `provider` + `model`. Three **families**: Galaxy, Ocean, Landmass.
The runtime builds one of three pipelines. Prefer the **situation stamp** table
for create defaults; use need-based rows for overrides.
`cascade` is current; never author legacy `cascade-01`. Never invent a
`tts_voice_id`. Galaxy and Ocean both support SIP. The server enforces model
access: some plans cannot save Ocean models (`403`), and `super` Scenarios
require Landmass.
## 1. Realtime — model speaks
Native transport. Only `provider` + `model`. Do not set `tts_*` / `llm_*` /
`stt_*`.
```json
{ "provider": "Galaxy", "model": "medium-stable" }
```
```json
{ "provider": "Ocean", "model": "medium-stable" }
```
Landmass realtime (`super` Scenarios, live avatars, or other Landmass-only
features) — still no `tts_*`:
```json
{ "provider": "Landmass", "model": "gemini" }
```
Galaxy `*-stable*` → Landmass `gemini-stable`. Ocean `medium-stable` → Landmass
`openai`.
## 2. Realtime with separate TTS
Landmass realtime (`gemini`, `gemini-stable`, or `openai`) + `tts_*` when the
user needs a selected/cloned voice but still wants a realtime listener. Load
[../recipes/cascade-tts.md](../recipes/cascade-tts.md) **only** for full Cascade
— not this path.
```json
{
"provider": "Landmass",
"model": "gemini",
"tts_provider": "cartesia",
"tts_voice_id": "[SELECTED_VOICE_ID]"
}
```
## 3. Cascade — STT + LLM + TTS
Landmass `cascade` for full pipeline control, selected external voice, or
text-first speech. Set component fields when choosing deliberately; omitted ones
default to OpenAI TTS, a Google `gemini-3.1-flash-lite` LLM, and Deepgram STT.
Load [../recipes/cascade-tts.md](../recipes/cascade-tts.md) into
`ai_instructions`.
```json
{
"provider": "Landmass",
"model": "cascade",
"tts_provider": "cartesia",
"tts_voice_id": "[SELECTED_VOICE_ID]",
"llm_provider": "google_vertex",
"llm_model": "gemini-3.1-flash-lite",
"stt_provider": "deepgram"
}
```
`tts_provider`: `cartesia` | `openai` | `elevenlabs` | `sarvam`.\
`llm_provider`: `google` | `google_vertex` | `openai` | `cerebras` | `sarvam`.\
`stt_provider`: `deepgram` | `cartesia` | `sarvam` | `google`. `stt_keyterms`:
terms to boost Deepgram recognition. `speed`: TTS rate multiplier (0.25-4;
providers clamp tighter).
`gemini_realtime_config` tunes Gemini Live (Landmass `gemini*` only):
`start_of_speech_sensitivity` / `end_of_speech_sensitivity` (`low`|`high`),
`silence_duration_ms`, `prefix_padding_ms`, `activity_handling`
(`interrupt`|`no_interruption`), `thinking_level`
(`minimal`|`low`|`medium`|`high`), `thinking_budget`, `proactivity`,
`enable_affective_dialog`. Unsupported settings are dropped server-side.
## Need-based overrides
| Need | Start with |
| ------------------------------------- | ------------------------------------- |
| Roleplay / multilingual chat | Galaxy `medium-stable` |
| Browser tools / interactive artifacts | Ocean `medium-stable` |
| Selected or cloned external voice | Landmass realtime + `tts_*` |
| Polished TTS / full STT·LLM·TTS | Landmass `cascade` |
| `super` Scenario / live avatar | Landmass `gemini` or `gemini-stable` |
| Silent meeting assistant | `meet_assist`; verify its live schema |
## Situation → starting stamp
Starting defaults:
| Situation | Stamp | Pipeline |
| ----------------- | ----------------------------- | -------- |
| Cold call / SDR | Landmass `cascade` + Cartesia | Cascade |
| Sales roleplay | Galaxy `medium-stable` | Realtime |
| Coaching | Ocean `medium-stable` | Realtime |
| Demo — browser | Ocean `medium-stable` | Realtime |
| Demo — slides | Landmass `cascade` + Cartesia | Cascade |
| Text-only | Ocean `medium-stable` | — |
| High-quality text | Ocean `medium-super` | — |
Cold call **and** slide demos use Cascade: outbound / narrated AI speech wants
controllable TTS (Cartesia) + explicit STT/LLM. Load
[../recipes/cascade-tts.md](../recipes/cascade-tts.md) for both.
Overrides:
- Cold call → Ocean only if Cascade / Cartesia is unavailable.
- Cold call that is a long conversation with many back-and-forths → Cascade is
still the default; ask the user whether to try voice-to-voice (Landmass
`gemini-stable` or Galaxy). Results vary by team, so it is their call.
- Coaching → Galaxy if Ocean is unavailable.
- Sales also needs `skip_auto_start: true` (see [control.md](control.md)).
## Cartesia `tts_voice_id` quick-reference
| Persona | Locale | `tts_voice_id` |
| --------------- | ------ | -------------------------------------- |
| Indian female | en-IN | `343e5d21-9db1-4efa-a1bc-196777cce1bb` |
| American female | en-US | `f786b574-daa5-4673-aa0c-cbe3e8534c02` |
Use these or a voice ID the user supplies; never invent one. Native Galaxy/Ocean
voices live in `appearance.voice` ([control.md](control.md)): `Aoede`, `Puck`,
`Kore`, `Charon`, `Fenrir`, …
## Catalog families
- Galaxy: `medium`, `medium-stable`, `medium-super`, and noise-cancelling
`medium-nc`, `medium-stable-nc`
- Ocean: `medium-stable`, `medium-super`
- Landmass: `gemini`, `gemini-stable`, `openai`, `cascade`
Other codes may appear on older Scenarios (`cascade-01`, `medium-preview`,
`openai-1`, `openai-1.5`, `gemini-preview`); they are legacy or internal — do
not author them.
Tool schemas and server validation remain authoritative.
## Key Files
- [control.md](control.md) — strategy, tools, appearance, flags
- [../recipes/cascade-tts.md](../recipes/cascade-tts.md) — Cascade speech rules
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |