@vellumai/assistant 0.4.44 → 0.4.45

package/src/config/bundled-skills/twitter/SKILL.md

@@ -1,206 +0,0 @@
----
-name: twitter
-description: Read and post on X (formerly Twitter) via OAuth or managed mode
-compatibility: "Designed for Vellum personal assistants"
-metadata: {"emoji":"𝕏","vellum":{"display-name":"X","user-invocable":true}}
----
-
-You are an X (formerly Twitter) assistant. Use the `bash` tool to run `assistant x`, `assistant config`, and `assistant oauth` CLI commands.
-
-## Connection Options
-
-There are two supported ways to connect to X. Choose whichever fits the user's situation.
-
-### Managed mode (platform-hosted credentials)
-
-When `twitter.integrationMode` is set to `managed`, the platform holds the OAuth credentials and proxies Twitter API calls on behalf of the assistant. No local OAuth setup is needed.
-
-- Supports: **post**, **reply**, and **read** operations (routed through the platform proxy)
-- Prerequisites: The assistant must be registered with the platform (`PLATFORM_ASSISTANT_ID`), have an API key (`credential:vellum:assistant_api_key`), and the assistant owner must have connected their Twitter account on the platform.
-
-**Error scenarios in managed mode:**
-- *"Assistant not bootstrapped"* -- The assistant API key is missing. Run setup.
-- *"Local assistant not registered with platform"* -- `PLATFORM_ASSISTANT_ID` is not set.
-- *"Connect Twitter in Settings as the assistant owner"* -- The owner hasn't connected their X account on the platform yet.
-- *"Sign in as the assistant owner"* -- The current user is not the assistant owner.
-- *"Reconnect Twitter or retry"* -- The platform's OAuth token may have expired. Reconnect on the platform.
-
-**Architecture notes for managed mode:**
-
-- **Assistant hosting mode and Twitter credential mode are separate concepts.** An assistant can be self-hosted (local daemon) yet use managed Twitter credentials, or platform-hosted yet use local BYO OAuth. The `twitter.integrationMode` config controls credential mode; the assistant's hosting mode is determined by its lockfile entry.
-- **Managed Twitter is bound to the assistant owner.** Only the owner of the assistant (as determined by the platform) can connect or disconnect the Twitter account. Non-owner users receive a `403` with an `owner_only` or `owner_credential_required` error code.
-- **Connect/disconnect/status uses desktop authentication.** The macOS Settings UI calls the platform's Twitter OAuth endpoints using the user's WorkOS-issued token. This authenticates the human user, not the assistant.
-- **Actual Twitter API calls use assistant-level API key authentication.** At runtime, the proxy client sends `Authorization: Api-Key {assistant_api_key}` -- it never includes user-level tokens. This ensures the assistant's identity is what the platform uses for token lookup and rate limiting.
-- **The platform proxy handles token storage and refresh.** OAuth tokens are stored server-side by the platform. The assistant never sees or stores the Twitter OAuth access/refresh tokens in managed mode. Token refresh is handled transparently by the proxy.
-- **The daemon auth handler never starts local OAuth in managed mode.** When `integrationMode` is `managed`, `handleTwitterAuthStart` returns a managed-specific error code (`managed_auth_via_platform` or `managed_missing_api_key`) and never calls `orchestrateOAuthConnect`. This is a critical guardrail to prevent credential confusion.
-
-### OAuth (recommended with X developer credentials)
-
-OAuth uses the official X API v2. It is the most reliable local connection method.
-
-- Supports: **post** and **reply**
-- Setup: Collect the OAuth Client ID (and optional Client Secret) from the user in chat using `credential_store` with `action: "prompt"` (canonical field names: `client_id`, `client_secret`), then initiate the `twitter_auth_start` flow. See the **First-Use Decision Flow** for the full sequence.
-
-## First-Use Decision Flow
-
-When the user triggers a Twitter operation and no connection has been configured yet, follow these steps:
-
-1. **Check current status:**
-
-   ```bash
-   # Check if OAuth token is available
-   assistant oauth token twitter --json
-
-   # Check integration mode
-   assistant config get twitter.integrationMode
-   ```
-
-2. **Determine the best path:**
-   - If managed mode prerequisites are met (integration mode is `managed`, assistant is registered with the platform, API key is present, and the owner has connected their Twitter account), use managed mode.
-   - Otherwise, guide the user through OAuth setup.
-
-3. **Execute setup for the chosen path:**
-   - If managed: Confirm prerequisites are satisfied. Managed mode requires no local setup beyond platform registration.
-   - If OAuth: Collect the credentials in-chat using the secure credential prompt, then connect. Follow the **OAuth Setup Sequence** below.
-
-### OAuth Setup Sequence
-
-When the user chooses OAuth, collect their X developer credentials conversationally using the secure UI. The OAuth flow delegates to the generic connect orchestrator, which resolves the Twitter provider profile, computes scopes via policy, opens the X authorization page, verifies the user's identity, and stores tokens. The orchestrator also manages stale refresh-token cleanup and enforces integration-mode guards.
-
-1. **Collect the Client ID securely:**
-   Call `credential_store` with `action: "prompt"`, `service: "integration:twitter"`, `field: "client_id"`, `label: "X (Twitter) OAuth Client ID"`, `description: "Enter the Client ID from your X Developer App"`, and `placeholder: "your-client-id"`.
-
-2. **Collect the Client Secret (if applicable):**
-   Ask the user if their X app uses a confidential client (has a Client Secret). If yes, call `credential_store` with `action: "prompt"`, `service: "integration:twitter"`, `field: "client_secret"`, `label: "X (Twitter) OAuth Client Secret"`, `description: "Enter the Client Secret from your X Developer App (leave blank if using a public client)"`, and `placeholder: "your-client-secret"`.
-
-3. **Initiate the OAuth flow:**
-   Trigger the Twitter auth start flow. The connect orchestrator resolves the Twitter provider profile, computes scopes via policy, opens the X authorization page, verifies the user's identity, and stores tokens. Wait for the auth result.
-
-4. **Confirm success:**
-   Tell the user: "Great, your X account is connected! You can always update these credentials from the Settings page."
-
-## Failure Recovery Flow
-
-When a Twitter operation fails, follow these steps:
-
-1. **Detect the failure type from the error output:**
-   - `OAuth is not configured` -- the user chose OAuth but credentials are not set up.
-   - `Twitter API error (401)` -- OAuth token may be expired or revoked.
-   - `Cannot connect to assistant` -- the Vellum assistant is not running.
-   - `proxyErrorCode: "owner_credential_required"` -- managed mode: the assistant owner has not connected their X account on the platform.
-   - `proxyErrorCode: "owner_only"` -- managed mode: the current user is not the assistant owner.
-   - `proxyErrorCode: "auth_failure"` or `"upstream_failure"` -- managed mode: platform token issue, reconnect Twitter on the platform.
-   - `proxyErrorCode: "missing_assistant_api_key"` -- managed mode: the assistant is not bootstrapped.
-   - `proxyErrorCode: "missing_platform_assistant_id"` -- managed mode: the assistant is not registered with the platform.
-
-2. **Explain the likely cause clearly** to the user.
-
-3. **Suggest recovery steps:**
-   - If OAuth failed or is not configured: suggest reconnecting OAuth credentials or checking that the X developer app credentials are correct.
-   - If managed mode failed with a credential or ownership error: explain the specific issue and guide the user to resolve it on the platform (connect Twitter, sign in as owner, etc.).
-
-## Posting
-
-Before posting, check the integration mode:
-
-```bash
-MODE=$(assistant config get twitter.integrationMode)
-```
-
-Then post based on the mode:
-
-```bash
-# Managed mode -- route through platform proxy (no local token needed):
-assistant x post "The post text here" --managed
-
-# With OAuth token:
-TOKEN=$(assistant oauth token twitter)
-assistant x post "The post text here" --oauth-token "$TOKEN"
-```
-
-When `twitter.integrationMode` is `managed`, always use `--managed`. The platform proxy handles authentication.
-
-Returns JSON with `ok`, `tweetId`, `text`, `url`, and `pathUsed` fields. Share the URL with the user so they can verify the post. For managed mode errors, the response includes `proxyErrorCode` and `retryable` fields.
-
-## Replying
-
-Same setup as posting -- check integration mode first, then:
-
-```bash
-# Managed mode:
-assistant x reply <tweetUrl> "The reply text here" --managed
-
-# Local OAuth:
-TOKEN=$(assistant oauth token twitter)
-assistant x reply <tweetUrl> "The reply text here" --oauth-token "$TOKEN"
-```
-
-The first argument is a tweet URL (e.g. `https://x.com/user/status/123456`) or a bare tweet ID.
-
-## Reading
-
-Read operations are available in managed mode. They route through the platform proxy.
-
-### User timeline
-
-```bash
-assistant x timeline <screenName> [--count N]
-```
-
-Returns `user` and `tweets` array.
-
-### Single tweet + replies
-
-```bash
-assistant x tweet <tweetIdOrUrl>
-```
-
-Returns the focal tweet and its reply thread.
-
-### Search
-
-```bash
-assistant x search "query" [--product Top|Latest|People|Media]
-```
-
-## Workflows
-
-### Check Mentions
-
-When the user asks to check mentions, check X, or see what's happening:
-
-1. Fetch their recent tweets to see replies: `assistant x timeline <theirScreenName> --count 10 --json`
-2. Summarize what needs attention:
-   - Group by type: replies to their tweets, mentions
-   - For anything that looks like it needs a reply, fetch the full thread with `assistant x tweet <tweetId>` to understand context
-   - Prioritize: direct questions > mentions > engagement
-3. For items that need replies, draft a response and ask the user to approve before sending with `assistant x reply`
-
-Present the summary as a scannable list, not a wall of text. Lead with action items.
-
-### Research a Topic
-
-When the user wants to understand what people are saying about something:
-
-1. Search: `assistant x search "topic" --count 20 --json`
-2. For the most interesting tweets, fetch threads: `assistant x tweet <tweetId>`
-3. Summarize: key themes, notable voices, sentiment, and any emerging consensus
-4. If the user wants to engage, draft a post or reply that adds to the conversation
-
-### Engagement Check
-
-When the user wants to see how their posts are performing:
-
-1. Fetch their recent tweets: `assistant x timeline <screenName> --count 20 --json`
-2. For each tweet, note engagement signals from the text/metadata
-3. Summarize: which posts got traction, who's engaging, any conversations worth continuing
-
-## Tips
-
-- Keep posts under 280 characters
-- All `screenName` arguments should be without the `@` prefix
-- All commands return JSON with an `ok` field
-- When drafting replies, match the tone of the conversation -- casual threads get casual replies
-- Always show the user what you're about to post and get approval before sending
-- If an operation fails, check `assistant x status --json` to diagnose the issue before retrying
-- The `post` and `reply` commands include a `pathUsed` field in their response so you can tell the user which connection method was used

package/src/config/bundled-skills/twitter/icon.svg

@@ -1,14 +0,0 @@
-<svg viewBox="0 0 16 16" xmlns="http://www.w3.org/2000/svg">
-<rect width="16" height="16" fill="#ffffff"/>
-<rect x="2" y="2" width="12" height="12" fill="#000000"/>
-<rect x="4" y="5" width="2" height="1" fill="#ffffff"/>
-<rect x="5" y="4" width="1" height="3" fill="#ffffff"/>
-<rect x="6" y="5" width="2" height="1" fill="#ffffff"/>
-<rect x="10" y="5" width="2" height="1" fill="#ffffff"/>
-<rect x="11" y="4" width="1" height="3" fill="#ffffff"/>
-<rect x="12" y="5" width="2" height="1" fill="#ffffff"/>
-<rect x="4" y="9" width="1" height="2" fill="#ffffff"/>
-<rect x="5" y="8" width="6" height="1" fill="#ffffff"/>
-<rect x="11" y="9" width="1" height="2" fill="#ffffff"/>
-<rect x="5" y="11" width="6" height="1" fill="#ffffff"/>
-</svg>

package/src/config/bundled-skills/typescript-eval/SKILL.md

@@ -1,60 +0,0 @@
----
-name: typescript-eval
-description: Test TypeScript code snippets before persisting as skills
-compatibility: "Designed for Vellum personal assistants"
-metadata: {"emoji":"🧪","vellum":{"display-name":"TypeScript Evaluation","user-invocable":false}}
----
-
-# TypeScript Evaluation
-
-When you need to test a TypeScript snippet before persisting it as a managed skill, use `bash` directly.
-
-## Workflow
-
-### 1. Write the snippet to a temp file
-
-```
-bash command="mkdir -p /tmp/vellum-eval && cat > /tmp/vellum-eval/snippet.ts << 'SNIPPET_EOF'
-<your code here>
-SNIPPET_EOF"
-```
-
-### 2. Run it with bun
-
-```
-bash command="bun run /tmp/vellum-eval/snippet.ts" timeout_seconds=10
-```
-
-### 3. For function-based testing
-
-If the snippet exports a `default` or `run` function, write a runner script:
-
-```
-bash command="cat > /tmp/vellum-eval/runner.ts << 'RUNNER_EOF'
-import * as mod from './snippet.ts';
-const fn = (mod as any).default ?? (mod as any).run;
-const input = {}; // mock input
-const result = await fn(input);
-console.log(JSON.stringify(result, null, 2));
-RUNNER_EOF"
-```
-
-Then run the runner:
-
-```
-bash command="bun run /tmp/vellum-eval/runner.ts" timeout_seconds=10
-```
-
-### 4. Clean up
-
-```
-bash command="rm -rf /tmp/vellum-eval/"
-```
-
-## Guidelines
-
-- **Iteration limit:** Max 3 attempts before asking the user for guidance.
-- **After successful test:** Persist with `scaffold_managed_skill` only after explicit user consent.
-- **Timeout:** Use `timeout_seconds=10` (or up to 20 for complex snippets).
-- **Error handling:** Read stdout/stderr from the bash output to diagnose failures.
-- **Never persist or delete skills without explicit user confirmation.**

package/src/config/bundled-skills/vercel-token-setup/SKILL.md

@@ -1,214 +0,0 @@
----
-name: vercel-token-setup
-description: Set up a Vercel API token for publishing apps using browser automation
-compatibility: "Designed for Vellum personal assistants"
-metadata: {"emoji":"▲","vellum":{"display-name":"Vercel Token Setup","includes":["browser"],"credential-setup-for":"vercel:api_token"}}
----
-
-You are helping your user set up a Vercel API token so they can publish apps to the web.
-
-## Client Check
-
-Determine whether the user has browser automation available (macOS desktop app) or is on a non-interactive channel (Telegram, SMS, etc.).
-
-- **macOS desktop app**: Follow the **Automated Setup** path below.
-- **Telegram or other channel** (no browser automation): Follow the **Manual Setup for Channels** path below.
-
----
-
-# Path A: Manual Setup for Channels (Telegram, SMS, etc.)
-
-When the user is on Telegram or any non-macOS client, walk them through a text-based setup. No browser automation is used — the user follows links and performs each action manually.
-
-### Channel Step 1: Confirm and Explain
-
-Tell the user:
-
-> **Setting up Vercel API Token**
->
-> Since I can't automate the browser from here, I'll walk you through each step with direct links. You'll need:
-> 1. A Vercel account (free tier works)
-> 2. About 2 minutes
->
-> Ready to start?
-
-If the user declines, acknowledge and stop.
-
-### Channel Step 2: Create the Token
-
-Tell the user:
-
-> **Step 1: Create an API token**
->
-> Open this link to go to your Vercel tokens page:
-> https://vercel.com/account/tokens
->
-> 1. Click **"Create"** (or **"Create Token"**)
-> 2. Set the token name to **"Vellum Assistant"**
-> 3. Select scope: **"Full Account"**
-> 4. Set expiration to the longest option available (or **"No Expiration"** if offered)
-> 5. Click **"Create Token"**
->
-> A token value will appear — **copy it now**, as it's only shown once.
-
-### Channel Step 3: Store the Token
-
-Tell the user:
-
-> **Step 2: Send me the token**
->
-> Please paste the token value into the secure prompt below.
-
-Present the secure prompt:
-
-```
-credential_store prompt:
-  service: "vercel"
-  field: "api_token"
-  label: "Vercel API Token"
-  description: "Paste the API token you just created on vercel.com"
-  placeholder: "Enter your Vercel API token"
-```
-
-Wait for the user to complete the prompt. Once received, store it:
-
-```
-credential_store store:
-  service: "vercel"
-  field: "api_token"
-  value: "<the token the user provided>"
-  allowedTools: ["publish_page", "unpublish_page"]
-  allowedDomains: ["api.vercel.com"]
-```
-
-### Channel Step 4: Done!
-
-> **Vercel is connected!** You can now publish apps to the web. Try clicking Publish on any app you've built.
-
----
-
-# Path B: Automated Setup (macOS Desktop App)
-
-You will automate Vercel token creation via the browser while the user watches. The user's only manual action is signing in to Vercel (if needed) and one copy-paste for the token value.
-
-## Browser Interaction Principles
-
-Vercel's UI may change over time. Do NOT memorize or depend on specific element IDs, CSS selectors, or DOM structures. Instead:
-
-1. **Screenshot first, act second.** Before every interaction, take a `browser_screenshot` to see the current visual state. Use `browser_snapshot` to find interactive elements.
-2. **Adapt to what you see.** If a button's label or position differs from what you expect, use the screenshot to find the correct element.
-3. **Verify after every action.** After clicking, typing, or navigating, take a new screenshot to confirm the action succeeded.
-4. **Never assume DOM structure.** Use the snapshot to identify what's on the page and interact accordingly.
-5. **When stuck, screenshot and describe.** If you cannot find an expected element after 2 attempts, take a screenshot, describe what you see to the user, and ask for guidance.
-
-## Anti-Loop Guardrails
-
-Each step has a **retry budget of 3 attempts**. An attempt is one try at the step's primary action (e.g., clicking a button, filling a form). If a step fails after 3 attempts:
-
-1. **Stop trying.** Do not continue retrying the same approach.
-2. **Fall back to manual.** Tell the user what you were trying to do and ask them to complete that step manually in the browser. Give them the direct URL and clear text instructions.
-3. **Resume automation** at the next step once the user confirms the manual step is done.
-
-If **two or more steps** require manual fallback, abandon the automated flow entirely and switch to giving the user the remaining steps as clear text instructions with links.
-
-## Things That Do Not Work — Do Not Attempt
-
-These actions are technically impossible in the browser automation environment:
-
-- **Downloading files.** `browser_click` on a Download button does not save files to disk.
-- **Reading the token value from a screenshot.** The token IS visible in the creation dialog, but you MUST NOT attempt to read it from a screenshot — it is too easy to misread characters, and the value must be exact. Always use the `credential_store prompt` approach to let the user copy-paste it accurately.
-- **Clipboard operations.** You cannot copy/paste via browser automation.
-
-## Step 1: Single Upfront Confirmation
-
-Tell the user:
-
-> **Setting up your Vercel API token so we can publish your app...**
->
-> Here's what will happen:
-> 1. **A browser opens** to your Vercel account settings
-> 2. **You sign in** (if not already signed in)
-> 3. **I create the token** — you just watch
-> 4. **One quick copy-paste** — I'll ask you to copy the token value into a secure prompt
->
-> Takes about a minute. Ready?
-
-If the user declines, acknowledge and stop. No further confirmations are needed after this point.
-
-## Step 2: Open Vercel and Sign In
-
-**Goal:** The user is signed in and the Vercel tokens page is loaded.
-
-Navigate to `https://vercel.com/account/tokens`.
-
-Take a screenshot and snapshot to check the page state:
-- **Sign-in page:** Tell the user: "Please sign in to your Vercel account in the browser." Then auto-detect sign-in completion by polling screenshots every 5-10 seconds. Check if the current URL has moved away from the login/sign-in page to the tokens page. Do NOT ask the user to "let me know when you're done" — detect it automatically. Once sign-in is detected, tell the user: "Signed in! Creating your API token now..."
-- **Already signed in:** Tell the user: "Already signed in — creating your API token now..." and continue immediately.
-
-**Verify:** URL contains `vercel.com/account/tokens` and no sign-in overlay is visible.
-
-## Step 3: Create Token
-
-**Goal:** A new API token named "Vellum Assistant" is created.
-
-Take a screenshot and snapshot. Find and click the button to create a new token (typically labeled "Create" or "Create Token").
-
-On the creation form:
-- Token name: **"Vellum Assistant"**
-- Scope: Select **"Full Account"** (or the broadest scope available)
-- Expiration: Select the longest option available, or **"No Expiration"** if offered
-- Click create/submit
-
-**Verify:** Take a screenshot. A dialog or section should now display the newly created token value.
-
-## Step 4: Capture Token via Secure Prompt
-
-**Goal:** The token value is securely captured and stored.
-
-### CRITICAL — Token Capture Protocol
-
-After token creation, Vercel shows the token value **once**. You MUST follow this exact sequence — **no improvisation**:
-
-1. Tell the user: "Your token has been created! Please copy the token value shown on screen and paste it into the secure prompt below."
-2. **IMMEDIATELY** present a `credential_store prompt` for the token. This is your ONLY next action.
-3. Wait for the user to paste the token.
-
-**Absolute prohibitions during this step:**
-- Do NOT try to read the token value from the screenshot. It must come from the user via secure prompt to ensure accuracy.
-- Do NOT navigate away from the page until the user has pasted the token.
-- Do NOT click any download or copy buttons.
-
-Present the secure prompt:
-
-```
-credential_store prompt:
-  service: "vercel"
-  field: "api_token"
-  label: "Vercel API Token"
-  description: "Copy the token value shown on the Vercel page and paste it here."
-  placeholder: "Enter your Vercel API token"
-```
-
-Wait for the user to complete the prompt. Once received, store it:
-
-```
-credential_store store:
-  service: "vercel"
-  field: "api_token"
-  value: "<the token the user provided>"
-  allowedTools: ["publish_page", "unpublish_page"]
-  allowedDomains: ["api.vercel.com"]
-```
-
-**Verify:** `credential_store list` shows `api_token` for `vercel`.
-
-## Step 5: Done!
-
-"**Vercel is connected!** Your API token is set up and ready to go. You can now publish apps to the web."
-
-## Error Handling
-
-- **Page load failures:** Retry navigation once. If it still fails, tell the user and ask them to check their internet connection.
-- **Element not found:** Take a fresh screenshot to re-assess. The Vercel UI may have changed. Describe what you see and try alternative approaches. If stuck after 2 attempts, ask the user for guidance.
-- **Token already exists with same name:** This is fine — Vercel allows multiple tokens with the same name. Proceed with creation.
-- **Any unexpected state:** Take a `browser_screenshot`, describe what you see, and ask the user for guidance.

package/src/config/bundled-skills/voice-setup/SKILL.md

@@ -1,131 +0,0 @@
----
-name: voice-setup
-description: Complete voice configuration in chat — PTT key, wake word, microphone permissions, ElevenLabs TTS, and troubleshooting
-compatibility: "Designed for Vellum personal assistants"
-metadata: {"emoji":"🎙️","vellum":{"display-name":"Voice Setup","user-invocable":true,"includes":["elevenlabs-voice"],"os":["darwin"]}}
----
-
-You are helping the user set up and troubleshoot voice features (push-to-talk, wake word, text-to-speech) entirely within this conversation. Do NOT direct the user to the Settings page for initial setup — handle everything in-chat using the tools below.
-
-## Available Tools
-
-- `voice_config_update` — Change any voice setting (PTT key, wake word enabled/keyword/timeout, TTS voice ID)
-- `open_system_settings` — Open macOS System Settings to a specific privacy pane
-- `navigate_settings_tab` — Open the Vellum settings panel to the Voice tab
-- `credential_store` — Collect API keys securely (for ElevenLabs TTS)
-
-## Setup Flow
-
-Walk the user through each section in order. Skip sections they don't need. Ask before proceeding to the next section.
-
-### 1. Microphone Permission
-
-Check `<channel_capabilities>` for `microphone_permission_granted`.
-
-**If `false` or missing:**
-
-1. Explain that macOS requires microphone permission for voice features.
-2. Use `open_system_settings` with `pane: "microphone"` to open the right System Settings pane.
-3. Tell the user: "I've opened System Settings to the Microphone section. Please toggle **Vellum Assistant** on, then come back here."
-4. After they confirm, verify by checking capabilities on the next turn.
-
-**If `true`:** Tell them microphone is already granted and move on.
-
-### 2. Push-to-Talk Activation Key
-
-Present common PTT key options:
-
-- **Right Option** — Default, good general choice
-- **Fn** — Dedicated key on most Mac keyboards
-- **Right Command** — Easy to reach
-- **Right Control** — Familiar from gaming
-
-Ask which key they prefer, then use `voice_config_update` with `setting: "activation_key"` and the chosen value.
-
-**Common issues to mention:**
-
-- If they pick a key that conflicts with their emoji picker (Fn or Globe on newer Macs), warn them and suggest an alternative.
-- If they use a terminal app heavily, warn that some keys may be captured by the terminal.
-
-### 3. Wake Word (Optional)
-
-Ask if they want to enable wake word detection (hands-free activation by saying a keyword).
-
-**If yes:**
-
-1. Use `voice_config_update` with `setting: "wake_word_enabled"`, `value: true`.
-2. Ask what wake word they want. Common choices: "Hey Vellum", "Computer", "Jarvis", their assistant's name.
-3. Use `voice_config_update` with `setting: "wake_word_keyword"` and their chosen word.
-4. Ask about timeout (how long the mic stays active after wake word). Options: 5s, 10s (default), 15s, 30s, 60s.
-5. Use `voice_config_update` with `setting: "wake_word_timeout"` and their chosen value.
-
-**Speech Recognition permission:** Wake word requires Speech Recognition access. Check capabilities — if not granted, use `open_system_settings` with `pane: "speech_recognition"`.
-
-### 4. Text-to-Speech / ElevenLabs (Optional)
-
-Ask if they want high-quality text-to-speech voices via ElevenLabs (optional — standard TTS works without it).
-
-If yes, the included **ElevenLabs Voice** skill (automatically appended below via `includes`) provides the full setup flow: curated voice list, API key collection, advanced voice selection, and tuning parameters. Follow the instructions there.
-
-Note: The shared config key `elevenlabs.voiceId` controls the voice for both in-app TTS and phone calls. If the user sets up phone calls later, they will automatically use the same voice for a consistent experience.
-
-### 5. Verification
-
-After setup is complete:
-
-1. Summarize what was configured.
-2. Suggest they test by pressing their PTT key (or saying their wake word) and speaking.
-3. Offer to open the Voice settings tab if they want to review: use `navigate_settings_tab` with `tab: "Voice"`.
-
-## Troubleshooting Decision Trees
-
-When the user reports a problem, follow the appropriate decision tree:
-
-### "PTT isn't working" / "Can't record"
-
-1. **Microphone permission** — Check `microphone_permission_granted` in capabilities. If false, guide through granting it.
-2. **Key check** — Ask what key they're using. Confirm it matches their configured PTT key.
-3. **Emoji picker conflict** — On newer Macs, Fn/Globe opens the emoji picker. If they're using Fn, suggest switching to Right Option or Right Command.
-4. **Speech Recognition permission** — Some voice features need this. Use `open_system_settings` with `pane: "speech_recognition"`.
-5. **App focus** — PTT may not work when Vellum is not the frontmost app or if another app has captured the key.
-
-### "Recording but no text" / "Transcription not working"
-
-1. **Speech Recognition permission** — Must be granted for transcription.
-2. **Microphone input** — Ask if they see the recording indicator. If yes, the mic works but transcription is failing.
-3. **Locale/language** — Speech recognition works best with the system language. Ask if they're speaking in a different language.
-4. **Background noise** — Excessive noise can prevent transcription. Suggest a quieter environment or a closer microphone.
-
-### "Wake word not detecting"
-
-1. **Enabled check** — Confirm wake word is enabled in their settings.
-2. **Keyword** — Confirm what keyword they're using. Shorter or common words may have lower accuracy.
-3. **Ambient noise** — Wake word detection is sensitive to background noise.
-4. **Permissions** — Both Microphone and Speech Recognition permissions are required.
-5. **Timeout** — If wake word activates but cuts off too quickly, increase the timeout.
-
-### "Changed a setting but it didn't work"
-
-1. **IPC broadcast** — The setting should take effect immediately. If it didn't, suggest restarting the assistant.
-2. **Verify** — Open the Voice settings tab with `navigate_settings_tab` to confirm the setting was persisted.
-
-## Deep Debugging
-
-For persistent issues, suggest checking system logs:
-
-```bash
-log stream --predicate 'subsystem == "com.vellum.assistant"' --level debug
-```
-
-Key log categories:
-
-- `voice` — PTT activation, recording state
-- `wake-word` — Wake word detection events
-- `speech` — Speech recognition results
-
-## Rules
-
-- Always handle setup conversationally in-chat. Do NOT tell the user to go to Settings for initial configuration.
-- Use `navigate_settings_tab` only for review/verification after in-chat setup, not as the primary setup method.
-- Be concise. Don't explain every option exhaustively — present the most common choices and let the user ask for more.
-- If a permission is denied, acknowledge it gracefully and explain what features won't work without it.

package/src/config/bundled-skills/voice-setup/icon.svg

@@ -1,20 +0,0 @@
-<svg viewBox="0 0 16 16" xmlns="http://www.w3.org/2000/svg">
-  <rect x="7" y="1" width="2" height="2" fill="#4A90E2"/>
-  <rect x="6" y="3" width="4" height="1" fill="#4A90E2"/>
-  <rect x="5" y="4" width="6" height="1" fill="#7B68EE"/>
-  <rect x="4" y="5" width="8" height="1" fill="#7B68EE"/>
-  <rect x="4" y="6" width="8" height="1" fill="#4A90E2"/>
-  <rect x="5" y="7" width="6" height="1" fill="#4A90E2"/>
-  <rect x="6" y="8" width="4" height="1" fill="#7B68EE"/>
-  <rect x="7" y="9" width="2" height="1" fill="#7B68EE"/>
-  <rect x="3" y="10" width="2" height="4" fill="#50C878"/>
-  <rect x="4" y="10" width="1" height="1" fill="#50C878"/>
-  <rect x="5" y="11" width="1" height="1" fill="#50C878"/>
-  <rect x="11" y="10" width="2" height="4" fill="#E74C3C"/>
-  <rect x="10" y="10" width="1" height="1" fill="#E74C3C"/>
-  <rect x="9" y="11" width="1" height="1" fill="#E74C3C"/>
-  <rect x="6" y="11" width="4" height="1" fill="#FFD700"/>
-  <rect x="5" y="12" width="6" height="1" fill="#FFD700"/>
-  <rect x="4" y="13" width="8" height="1" fill="#FFD700"/>
-  <rect x="7" y="14" width="2" height="2" fill="#4A90E2"/>
-</svg>