…iles

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Use product switcher for navigation and add icons to Atoms and Waves

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…lues

fix: correct output_format enum, add v3.1 voice discovery, and fix unified SDK overrides

Python SDK:
- Add additional_init_exports to export AtomsClient, AsyncAtomsClient,
  WavesClient, and AsyncWavesClient directly from the package root
- Users can now do: from smallest_ai import AtomsClient, WavesClient

TypeScript SDK:
- Enable generateSubpackageExports to allow importing individual clients
- Users can now do: import { AtomsClient } from 'smallest-ai/atoms'

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…eExports support

The previous version (2.6.3) did not support the generateSubpackageExports config option.

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…-exports

feat: enable additional init exports for Python and TypeScript SDKs

- Add environments configuration in unified generators.yml with separate base URLs:
  - atoms: https://atoms-api.smallest.ai/api/v1
  - waves: https://waves-api.smallest.ai
- Add x-fern-server-name: waves to all waves OpenAPI and AsyncAPI endpoints
- Create new override files for add-voice, get-cloned-voices, and delete-cloned-voice endpoints
- This fixes the issue where waves requests were incorrectly sent to atoms-api.smallest.ai

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…se-urls

fix: configure multiple base URLs for atoms and waves APIs

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…t-waves

Enable websocket client generation for waves API

- Updated Atoms content with new pages for intro, platform, developer guide, and API reference tabs
- Updated Waves content for v4.0.0, v3.0.1, and v2.2.0 with new documentation structure
- Added new images and videos for both Atoms and Waves documentation
- Fixed image paths in MDX files to use correct relative paths
- Updated navigation YAML files to match new content structure

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…ages

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

Co-Authored-By: blank@buildwithfern.com <blank@buildwithfern.com>

…ves-toggle

Update Atoms and Waves content with new documentation structure

…/images)

Made-with: Cursor

docs: sync content with atoms-platform — MCP tab, quick-start, new cookbooks

* update: docs and api spec with new URLs

* fix: TTS playground fix

* fix: hyperlink updating

* fix: broken URLs

The waves and atoms generators.yml files had SDK generator groups (python-sdk,
ts-sdk, go-sdk) with autorelease enabled, which conflicted with the unified
config. The waves config was generating to the same repos as unified but without
an explicit package_name, causing the Python generator to derive 'smallest_ai'
from the org name instead of using the intended 'smallestai'.

This removes the groups and autorelease sections while keeping the API
definitions intact (still referenced by the unified config via relative paths).

Co-Authored-By: blank <blank@buildwithfern.com>

…generators

fix: remove legacy SDK generator groups from waves and atoms configs

… Agents / Models

- Atoms Platform tab now appears first in navigation (before Product Overview)
- Product toggle renamed: Atoms → Voice Agents, Waves → Models
- Add availability: deprecated to Conversational Flow Agents section
- Fix all nav display names to Title Case matching Mintlify
- Add Campaigns overview, SSE Live Transcripts to Developer Guide
- Restructure Developer Guide: separate Operate section
- Add Reference section to Platform tab

Content sync:
- Sync 11 diverged pages from atoms-platform/docs (quick-start, welcome,
  intro, cookbooks, convo-flow overview, MCP pages, quickstart, etc.)
- Add campaigns/overview.mdx (was missing)
- Add 7 images (API keys screenshots, platform settings, etc.)

Link fixes:
- Rewrite all 69 internal links from Mintlify paths to Fern URL paths
- Fix Discord icons to fa-brands fa-discord
- Fix all image paths to relative paths

UX improvements:
- Add Deprecated badge to all Conversational Flow card references (9 cards)
- Rewrite Getting Help page with Discord, Showcase, Support CardGroups
- Add "Still Have Questions?" section to FAQ with Discord/Email/Showcase
- Fix Cookbook/GitHub icons to fa-brands fa-github

API:
- Add POST /product/import-phone-number (BYOSIP) endpoint to OpenAPI spec
- Update telephony.mdx with SIP import documentation

- Rename: How to TTS → Sync & Async, Stream TTS → Streaming
- Rename: Voice Models & Languages → Voices & Languages
- Rename: HTTP Streaming → HTTP vs Streaming vs WebSockets
- Add Model Cards as separate top-level tab
- Add Word Boosting to STT Features
- Remove Types of Cloning and Professional Voice Cloning from nav
- Remove PVC Best Practices from nav
- Add Vercel AI SDK and OpenClaw to Integrations
- TTS/STT sections: quickstart before overview
- Section names: Text to Speech (Lightning), Speech to Text (Pulse)

…ations

Content sync:
- Update 8 diverged pages with richer Mintlify content (quickstart, overview,
  how-to-tts, stream-tts, authentication, introduction, STT quickstart, etc.)
- Add 3 new pages: Lightning v3.1 model card, Pulse model card, TTS cookbook
- Add 2 new integration pages: Vercel AI SDK, OpenClaw
- Sync Vercel AI SDK and OpenClaw with latest branch updates

Assets:
- Add 2 audio files (tts-sample-hello.wav, stt-sample-audio.wav)
- Add 4 images (API keys screenshots)

Fixes:
- Models page: add Deprecated badge on v2, link cards to model cards,
  move Tip below STT section, fix card icons
- Fix all Discord icons to fa-brands fa-discord, GitHub to fa-brands fa-github
- Fix STT pre-recorded features card links (were giving 404s)
- Add deprecated badges to Lightning v2 API ref pages
- Convert Next Steps bullet lists to CardGroup on pronunciation dicts,
  best practices, voice cloning pages
- Fix TTS quickstart image alignment (consistent 700px width)
- Fix Voices & Languages page title
- Fix pvc-best-practices stale link
- Add word-boosting.mdx content

Adds the Context7 verification file required to claim the Smallest AI
library on Context7. This enables admin access to manage how our docs
are indexed for LLM-powered coding tools (Cursor, Claude, Copilot).

Made-with: Cursor

Adds a redirect in docs.yml so docs.smallest.ai/context7.json resolves
to the raw GitHub file, allowing Context7 to verify library ownership.

Made-with: Cursor

Add Context7 library claim verification

…hots

Broken link fixes:
- Fix analytics-and-logs → analytics-logs across 54 files (footer cards,
  Test Your Agent, Conversation Logs links were all 404)
- Fix "Behaviour" → "Behavior" spelling

Agent creation pages:
- Manual setup: restructure steps (Click Start From Scratch, Select Single
  Prompt, Configure Agent), add 4 new screenshots, add Test Your Agent section
- From template: fix step 1 title, add Test Your Agent section
- AI assisted: fix step 1 title, add Test Your Agent section
- Quick start: replace Step 3 screenshot with new create-with-ai-modal image
- All agent creation flows now end with consistent Test Your Agent step

Terminology:
- Replace "Waves TTS" → "Lightning TTS" across all pages
- Replace "Waves API" → "Smallest AI API" in content text (14 files)

Content fixes:
- Introduction: fix broken STT audio URL, add audio sample to TTS section,
  fix hello.wav description text
- Models: add keyword boosting mention to Pulse STT card
- Word boosting page: complete rewrite with keyword boosting documentation
  including format, intensifier scale, API examples, supported languages

Consistency:
- All deprecated badges use yellow color

Replace non-existent GitHub audio sample URL with a two-step approach:
generate audio via TTS API, then transcribe it. Verified both API calls
return 200 with correct responses.

… testing

- Manual setup: 6-step guide with Frame-captioned screenshots, configuration
  table, prompt writing guidance, 3 testing types (Web/Telephony/Chat)
- From template: 4-step guide with new screenshots, template customization
- AI assisted: mirrors quick-start flow, flat format (no accordions)
- Quick-start: wrap all images in Frame captions, update test section
- All pages now have consistent Test Your Agent section with 3 test types
- All images use Frame caption wrappers for proper labeling

Keyword boosting:
- Rewrite word-boosting.mdx with Real-Time badge, JS code example, Note callout
- Add keywords param to pulse-stt-ws AsyncAPI spec
- Add keywords to Pulse STT WebSocket API reference page (table + code)

Badge fixes:
- Models page: Lightning v2 Deprecated=orange, Lightning v3.1 Latest=green
- Both in cards and in Model Overview tables

Navigation:
- Move HTTP vs Streaming vs WebSockets from Getting Started → TTS section
- Fix Introduction page model card links (were 404)

Terminology:
- Replace "Waves TTS" → "Lightning TTS", "Waves API" → "Smallest AI API"

Broken links:
- Fix 26+ broken internal links across Atoms and Waves pages
- Glossary: troubleshooting/glossary → reference/glossary (3 files)
- Developer Guide overview: fix quickstart, core-concepts, analytics paths
- Waves STT overview: fix all pre-recorded/realtime paths to use pulse slug
- Waves TTS best practices: fix quickstart and voice cloning paths
- Fix voice-speech and welcome links in Atoms product overview pages

Content:
- Pulse model card: complete rewrite with 39 languages, streaming/non-streaming
  feature tables, best practices, limitations, use cases, safety
- Introduction page: Lightning/Pulse cards now link to model cards
- Move HTTP vs Streaming vs WebSockets to end of TTS section

API spec:
- Add full query bindings (including keywords param) to pulse-stt-ws
  override server. The override was stripping all query params because
  waves-ws server had no bindings, overriding the base spec's production
  server which did have them.

Feature pages:
- Realtime features: add Keyword Boosting card, fix all card links
  from /speech-to-text/ to /speech-to-text-pulse/ (were 404)
- Pre-recorded features: rename Word Boosting → Keyword Boosting
- Nav: rename Word Boosting → Keyword Boosting in sidebar

Add keyword boosting to Pulse STT API spec and fix feature pages

Remove the versions block from the Models (Waves) product config.
This eliminates the version dropdown (v4.0.0 / v3.0.1 / v2.2.0)
and serves only v4.0.0 content as the single version.

The v3.0.1 and v2.2.0 files are NOT deleted — they remain in the
repo in case we need to restore versioning later. They just won't
be served or shown in the UI.

URL impact:
- Before: /waves/v-4-0-0/documentation/... (with version prefix)
- After: /waves/documentation/... (no version prefix)