@vellumai/assistant 0.10.0 → 0.10.1-staging.1

package/src/config/bundled-skills/app-builder/references/INTERACTION_HOOKS.md

@@ -1,56 +1,83 @@
 # App Interaction Hooks
 
-When building apps, proactively wire `sendAction` hooks so the assistant stays aware of meaningful user interactions. Two patterns are available:
+A sandboxed app drives the assistant and its own on-screen layout through
+`window.vellum.sendAction(actionId, data)`. When the app is open inside a
+conversation — the chat panel, split view, or mobile app overlay, the surface
+this skill builds for — the host acts on exactly two actions: `relay_prompt`
+and `set_view`. Wire them during the build so the app can pull the assistant in
+and arrange itself.
 
-## Reactive hooks
+Both actions are about the app↔conversation relationship, so an app opened
+standalone — from the Library, with no chat beside it — has nothing listening,
+and `sendAction` there no-ops. Treat them as enhancements: every feature must
+still work when the call is ignored.
 
-Reactive hooks trigger an assistant response. Use them for moments where the assistant's input adds value - selections that need explanation, completions worth celebrating, or submissions that benefit from feedback.
+## `relay_prompt` — send a message to the assistant
 
-```javascript
-// User selects a city on a map — assistant can provide insights
-window.vellum.sendAction("city_selected", { city: "Tokyo" });
+Sends `prompt` into the conversation as if the user typed it; the assistant
+then responds in chat. This is the channel for involving the assistant from an
+app, so phrase the intent as natural language.
 
-// User submits a form — assistant can confirm and suggest next steps
-window.vellum.sendAction("form_submitted", {
-  formId: "signup",
-  email: "user@example.com",
+```javascript
+// A button in a dashboard: "Explain this anomaly"
+window.vellum.sendAction("relay_prompt", {
+  prompt: "Revenue dropped 40% on the 14th — what happened?",
 });
 
-// User completes a level — assistant can congratulate and hint at what's next
-window.vellum.sendAction("level_complete", { level: 5, score: 2400 });
+// Start a fresh conversation instead of using the open one
+window.vellum.sendAction("relay_prompt", {
+  prompt: "Draft a follow-up email to this customer.",
+  conversation: "new", // "active" (default) | "new"
+});
 ```
 
-## Silent hooks
+- **`prompt`** (required) — the message text. Empty or missing → ignored.
+- **`conversation`** — `"active"` (default: the open conversation) or
+  `"new"` (a fresh draft). With `"active"` and nothing open, it's a no-op.
+- **The layout is left exactly as-is.** Relaying never opens, closes, or
+  resizes the app — if the user has chat and app side by side, it stays that
+  way. Each relay is delivered even when the same text is sent repeatedly.
 
-Silent hooks accumulate state without interrupting the user. The state is automatically included as context when the next reactive hook fires.
+The assistant can't see the app's internal state, so make every prompt
+**self-contained**: include the numbers, selection, or row the user is acting
+on. "Summarize these" won't work; "Summarize these 3 expenses: $42 lunch,
+$118 hotel, $30 taxi" will.
 
-```javascript
-// User navigates to a new tab — no response needed, but assistant should know
-window.vellum.sendAction("state_update", {
-  currentView: "forecast",
-  city: "Tokyo",
-});
+Good triggers: "explain this chart," "what should I focus on," "turn these
+into tasks," "compare the two selected plans." Wire them on the buttons and
+selections where the assistant's help is the actual point.
 
-// Score changes during gameplay — track silently
-window.vellum.sendAction("state_update", { score: 1250, lives: 2 });
+### Triage / bulk-action UIs
 
-// User applies a filter — context for future questions
-window.vellum.sendAction("state_update", {
-  filter: "last-30-days",
-  sortBy: "revenue",
-});
+Render selectable items + an action button, then relay the user's choice as a
+prompt the assistant can act on:
+
+```javascript
+function runBulkAction(selected) {
+  const list = selected.map((s) => `- ${s.title} (${s.id})`).join("\n");
+  window.vellum.sendAction("relay_prompt", {
+    prompt: `Archive these ${selected.length} items:\n${list}`,
+  });
+}
 ```
 
-## When to use reactive vs silent
+## `set_view` — arrange the app and chat
+
+Moves the app panel. Independent of `relay_prompt`; use it only when the app
+itself should change the layout.
 
-Choose based on whether the assistant's response would genuinely help the user at that moment:
+```javascript
+window.vellum.sendAction("set_view", { view: "split" }); // app + chat side by side
+window.vellum.sendAction("set_view", { view: "full" }); // app full-width
+window.vellum.sendAction("set_view", { view: "chat" }); // close app, back to chat
+```
 
-| App type            | Silent (state accumulation)                          | Reactive (triggers response)                               |
-| ------------------- | ---------------------------------------------------- | ---------------------------------------------------------- |
-| **Dashboards**      | Tab navigation, filter changes, date range selection | Anomaly detected, threshold breached, data export complete |
-| **Games**           | Score updates, move tracking, timer ticks            | Level complete, achievement unlocked, game over            |
-| **Forms & wizards** | Field focus, partial input, step navigation          | Form submitted, validation failed on submit                |
-| **Trackers**        | Incremental progress, status toggles, reordering     | Milestone reached, streak achieved, all items complete     |
-| **Data explorers**  | Sorting, paging, column toggling                     | Row selected for detail, comparison initiated              |
+- **`"split"`** — side by side with the conversation. Desktop only; ignored on
+  mobile, which has no side-by-side layout.
+- **`"full"`** — the app takes the full width.
+- **`"chat"`** — closes the app and returns to the conversation.
 
-Wire hooks during the initial build - don't wait for the user to ask. Apps that communicate state back to the assistant feel alive; apps that don't feel like static pages.
+Most apps never need this — the user controls the layout from the app's nav
+bar. Reach for it only when an in-app action implies a layout change (e.g. a
+"Discuss with assistant" button that opens `split` so the reply lands beside
+the app).

package/src/config/bundled-skills/app-builder/references/RESPONSIVE.md

@@ -40,7 +40,7 @@ The conversation context's `<turn_context>` block carries an `interface:` field.
 
 - Use fluid widths only. No fixed-pixel layouts. Prefer `%`, `fr`, `minmax`, `clamp()` over `px` on container widths.
 - At narrow widths, collapse tables into stacked cards with labels and values arranged vertically.
-- Size `vellum.widgets.*` chart containers in `vw` / `%`, not fixed `px`.
+- Size chart containers (`chart.js` canvas or inline SVG) in `vw` / `%`, not fixed `px`.
 
 ---
 

package/src/config/bundled-skills/app-builder/references/WIDGETS.md

@@ -1,6 +1,6 @@
 # Widget Component Library
 
-A CSS/JS widget library is auto-injected alongside the design system. Use these for standard UI patterns - skip them when custom HTML serves the user better.
+A CSS widget library is auto-injected alongside the design system — the `.v-*` classes below. Use these for standard UI patterns; skip them when custom HTML serves the user better. The classes are styling only; wire any interactive behavior with your own JS (see below).
 
 ## Layout widgets
 
@@ -18,7 +18,7 @@ A CSS/JS widget library is auto-injected alongside the design system. Use these
 | `.v-progress-bar` / `.v-progress-track` / `.v-progress-fill` | Horizontal progress                                            |
 | `.v-status-badge`                                            | Colored pill with dot (`.success`/`.error`/`.warning`/`.info`) |
 | `.v-stat-row` / `.v-stat`                                    | Horizontal label-value pairs                                   |
-| `.v-toast`                                                   | Notification banner - prefer `vellum.widgets.toast()`          |
+| `.v-toast`                                                   | Notification banner (show/hide with your own JS)              |
 | `.v-avatar-row`                                              | Contact/team display                                           |
 | `.v-tag-group`                                               | Wrapping tag row                                               |
 
@@ -48,78 +48,20 @@ A CSS/JS widget library is auto-injected alongside the design system. Use these
 | `.v-gradient-text`                               | Accent-colored gradient text                           |
 | `.v-animate-in`                                  | Staggered fade-in for children                         |
 
-## Widget JavaScript utilities
+## Interactive behavior — your own JS
 
-Interactive utilities at `window.vellum.widgets.*`:
+The `.v-*` classes are styling only. Wire behavior with standard web tools:
 
-### Charts
+- **Charts** → the bundleable `chart.js`, or hand-written inline SVG / CSS bars for tiny sparklines, sized to the container to avoid overflow.
+- **Notifications** → toggle the `.v-toast` class with your own JS.
+- **Table sort/filter, tabs, accordions, countdowns** → plain JS event handlers.
+- **Formatting** → `Intl.NumberFormat` / `Intl.DateTimeFormat`.
+- **Theme** → `@media (prefers-color-scheme: dark)` in CSS.
 
-Always use these instead of hand-coding SVG/CSS charts:
+The complete in-app JS API is `window.vellum.sendAction`, `window.vellum.fetch`, and `window.vellum.route` (see [`INTERACTION_HOOKS.md`](./INTERACTION_HOOKS.md)).
 
-```javascript
-vellum.widgets.sparkline("container-id", [10, 25, 15, 30], {
-  width: 200,
-  height: 40,
-  color: "var(--v-success)",
-  strokeWidth: 2,
-  fill: true,
-});
-vellum.widgets.barChart(
-  "container-id",
-  [
-    { label: "Jan", value: 120 },
-    { label: "Feb", value: 180, color: "var(--v-success)" },
-  ],
-  {
-    width: 400,
-    height: 200,
-    showLabels: true,
-    showValues: true,
-    horizontal: false,
-  },
-);
-vellum.widgets.lineChart(
-  "container-id",
-  [
-    { label: "Mon", value: 42 },
-    { label: "Tue", value: 58 },
-  ],
-  { width: 400, height: 200, showDots: true, showGrid: true, gridLines: 4 },
-);
-vellum.widgets.progressRing("container-id", 75, {
-  size: 100,
-  strokeWidth: 8,
-  color: "var(--v-success)",
-  label: "75%",
-});
-```
+## When to use the CSS widgets vs custom HTML
 
-### Data Formatting
-
-```javascript
-vellum.widgets.formatCurrency(1234.56, "USD"); // "$1,234.56"
-vellum.widgets.formatDate("2025-01-15", "relative"); // "3d ago"
-vellum.widgets.formatDate("2025-01-15", "short"); // "1/15/25"
-vellum.widgets.formatNumber(1234567, { compact: true }); // "1.2M"
-```
-
-### Interactive Behaviors
-
-```javascript
-vellum.widgets.sortTable("table-id"); // Wire th[data-sortable] click-to-sort
-vellum.widgets.filterTable("table-id", "input-id"); // Live text search
-vellum.widgets.tabs("tabs-id"); // Tab switching + keyboard nav
-vellum.widgets.accordion("accordion-id", { allowMultiple: true });
-vellum.widgets.multiSelect("table-id"); // Checkboxes + select-all
-vellum.widgets.toast("Saved!", "success", 4000); // Auto-dismiss notification
-vellum.widgets.countdown("timer-el", "2025-12-31T00:00:00Z", {
-  onComplete: () => {},
-});
-```
-
-## When to use widgets vs custom HTML
-
-- **Use widgets** for standard patterns - tables, metrics, timelines, notifications
-- **Use custom HTML** for novel or creative UIs - games, art tools, unique dashboards
-- **Mix freely** - widgets compose well together and with custom elements
-- **ALWAYS use `vellum.widgets.*` chart functions** instead of hand-coding SVG/CSS charts. They handle overflow clipping, bounds, scaling, and dark mode. Hand-coded charts break layouts.
+- **Use the CSS widget classes** for standard patterns — tables, metrics, timelines, notifications.
+- **Use custom HTML** for novel or creative UIs — games, art tools, unique dashboards.
+- **Mix freely** — the classes compose well together and with custom elements.

package/src/config/bundled-skills/app-builder/references/examples/README.md

@@ -12,6 +12,5 @@ the `frontend-design` skill).
 | [Habit Tracker](./habit-tracker.md)     | Full CRUD addressed by `id` query param          | `GET`, `POST`, `PATCH`, `DELETE` |
 | [Expense Tracker](./expense-tracker.md) | Create / read / delete + client-side aggregation | `GET`, `POST`, `DELETE`          |
 
-All three persist through a `routes/*.ts` handler — never through `window.vellum.data`, which
-is a desktop-only bridge that silently no-ops on web. See
+All three persist through a `routes/*.ts` handler. See
 [CUSTOM_ROUTES.md](../CUSTOM_ROUTES.md) for the full route handler reference.

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

@@ -218,16 +218,14 @@ Use this when the guardian wants to authorize a specific phone number to call th
 **Important**: The response includes a `voiceCode` field that is only returned at creation time and cannot be retrieved later. Extract and present it clearly.
 
 ```bash
-assistant contacts invites create --source-channel phone --contact-id "<contact_id>" --expected-external-user-id "<phone_E164>" --friend-name "<invitee_name>" --guardian-name "<guardian_name>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
+assistant contacts invites create --source-channel phone --contact-id "<contact_id>" --expected-external-user-id "<phone_E164>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
 ```
 
 Required flags:
 
 - `--source-channel` -- must be `phone`
-- `--contact-id` -- the ID of the contact this invite is for. Look up the contact first with `assistant contacts list`. New contacts must be added via the Contacts tab in the assistant web UI; you cannot create them here.
+- `--contact-id` -- the ID of the contact this invite is for. Look up the contact first with `assistant contacts list`. New contacts must be added via the Contacts tab in the assistant web UI; you cannot create them here. The contact's `displayName` is used to personalize the voice greeting; the guardian label is resolved at runtime.
 - `--expected-external-user-id` -- the invitee's phone number in E.164 format (e.g., `+15551234567`)
-- `--friend-name` -- the invitee's display name (e.g., "Mom", "Dr. Smith"). Used during the voice verification call to personalize the experience.
-- `--guardian-name` -- the guardian's display name (e.g., "Alex"). Used during the voice verification call so the invitee knows who invited them.
 
 Optional flags:
 
@@ -235,10 +233,9 @@ Optional flags:
 - `--expires-in-ms` -- expiration time in milliseconds from now (e.g., `86400000` for 24 hours). Defaults to 7 days if omitted.
 - `--note` -- a human-readable label for the invite (e.g., "For Mom", "Dr. Smith")
 
-The create response contains `{ ok: true, invite: { id, voiceCode, expectedExternalUserId, friendName, guardianName, ... } }`.
+The create response contains `{ ok: true, invite: { id, voiceCode, expectedExternalUserId, ... } }`.
 
 - `voiceCode` is the numeric code the invitee must enter and is only returned at creation time.
-- `friendName` and `guardianName` are echoed back in the response.
 - Voice invite responses do **not** include `token` or `share.url`. Do not try to build or send a deep link for voice invites.
 
 **Presenting to the guardian**: Give the guardian clear instructions to relay to the invitee:
@@ -264,7 +261,7 @@ If the user provides a phone number without the `+` country code prefix, ask the
 Use this when the guardian wants to invite someone to message the assistant via email. Email invites use a 6-digit code - the invitee sends the code to the assistant's email address to redeem access.
 
 ```bash
-assistant contacts invites create --source-channel email --contact-id "<contact_id>" --contact-name "<invitee_name>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
+assistant contacts invites create --source-channel email --contact-id "<contact_id>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
 ```
 
 Required flags:
@@ -295,7 +292,7 @@ If the assistant's email address is not available, tell the guardian they need t
 Use this when the guardian wants to invite someone to message the assistant on WhatsApp. WhatsApp invites use a 6-digit code - the invitee sends the code to the assistant's WhatsApp number to redeem access.
 
 ```bash
-assistant contacts invites create --source-channel whatsapp --contact-id "<contact_id>" --contact-name "<invitee_name>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
+assistant contacts invites create --source-channel whatsapp --contact-id "<contact_id>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
 ```
 
 Required flags:
@@ -328,7 +325,7 @@ If the assistant's WhatsApp integration is not configured at all (Meta WhatsApp
 Use this when the guardian wants to invite someone to message the assistant on Slack. Slack invites use a 6-digit code -- the invitee sends the code as a direct message to the assistant's Slack bot to redeem access.
 
 ```bash
-assistant contacts invites create --source-channel slack --contact-id "<contact_id>" --contact-name "<invitee_name>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
+assistant contacts invites create --source-channel slack --contact-id "<contact_id>" --max-uses 1 --note "<optional note, e.g. the person it is for>" --json
 ```
 
 Required flags:
@@ -451,8 +448,6 @@ Each channel has:
   - `contactId is required for create` -- when creating an invite, always pass `--contact-id`. Look up or create the contact first.
   - `expectedExternalUserId is required for voice invites` -- voice invites must include the invitee's phone number via `--expected-external-user-id`.
   - `expectedExternalUserId must be in E.164 format` -- the phone number must start with `+` followed by country code and number (e.g., `+15551234567`).
-  - `friendName is required for voice invites` -- voice invites must include the invitee's display name via `--friend-name`.
-  - `guardianName is required for voice invites` -- voice invites must include the guardian's display name via `--guardian-name`.
   - `Invite not found or already revoked` -- the invite ID may be invalid or the invite is already revoked.
 
 ## Tips
@@ -485,7 +480,7 @@ Each channel has:
 
 **"Revoke invite"** / **"Cancel invite link"** -- List invites to identify the target, confirm, then revoke with `assistant contacts invites revoke <invite_id> --json`.
 
-**"Create a voice invite for +15551234567"** -- Look up or create the contact first, then create a voice invite with `assistant contacts invites create --source-channel phone --contact-id <contact_id> --expected-external-user-id "+15551234567" --friend-name "<name>" --guardian-name "<name>"`. Present the invite code and instructions: the person must call from that number and enter the code.
+**"Create a voice invite for +15551234567"** -- Look up or create the contact first, then create a voice invite with `assistant contacts invites create --source-channel phone --contact-id <contact_id> --expected-external-user-id "+15551234567"`. Present the invite code and instructions: the person must call from that number and enter the code. The caller's name and the guardian's name resolve from the bound contact and the runtime guardian — no name flags needed.
 
 **"Let my mom call in"** / **"Invite someone by phone"** -- Ask for the phone number in E.164 format, look up or create the contact first, then create a voice invite with `assistant contacts invites create --source-channel phone --contact-id <contact_id>`, and present the code + calling instructions.
 

package/src/config/bundled-skills/messaging/tools/shared.ts

@@ -137,5 +137,8 @@ export async function getProviderConnection(
 ): Promise<OAuthConnection | undefined> {
   if (provider.resolveConnection) return provider.resolveConnection(account);
   if (await provider.isConnected?.()) return undefined;
-  return resolveOAuthConnection(provider.credentialService, { account });
+  return resolveOAuthConnection(provider.credentialService, {
+    account,
+    requiredScopes: provider.requiredScopes,
+  });
 }

package/src/config/bundled-skills/system-storage-cleanup/SKILL.md

@@ -0,0 +1,74 @@
+---
+name: system-storage-cleanup
+description: "Handle disk pressure, critically low storage, safe storage limits, and storage cleanup mode by inspecting disk usage and proposing safe cleanup steps."
+compatibility: "Designed for Vellum personal assistants"
+metadata:
+  emoji: "\U0001F9F9"
+  vellum:
+    display-name: "System Storage Cleanup"
+    category: "system"
+    activation-hints:
+      - "Critical disk usage or disk-pressure lock is blocking normal work"
+      - "The assistant is in storage cleanup mode or normal work is suspended due to low storage"
+      - "User asks to inspect large files or find what is using disk space"
+      - "User asks to free workspace storage safely"
+      - "User asks to delete caches, logs, temp files, Docker artifacts, or other storage safely"
+---
+
+You are operating under a critical storage cleanup contract. Your only goal is to free enough storage for the assistant to resume normal work without damaging user data.
+
+## Cleanup Contract
+
+Start by warning the user that storage is critically low and normal work is suspended until storage cleanup mode clears. Stay scoped to freeing storage until the disk-pressure lock clears or the guardian explicitly overrides it.
+
+Prefer foreground inspection with available cleanup-safe tools before any mutation. Identify both the target volume that is actually full and the workspace path before proposing deletions. Do not work on unrelated tasks, refactors, installs, upgrades, or product changes while the storage lock is active.
+
+Ask for explicit approval before deleting files, caches, logs, package caches, Docker artifacts, or any other data unless the user has already approved that exact action. Before asking, present each proposed deletion with:
+
+- Exact path or artifact name.
+- Estimated reclaimable size.
+- Expected consequence, including whether it is regenerable or may remove user-visible history.
+
+If the user approves a broad category, narrow it to exact paths or artifacts before deleting. If the user approves one exact path, do not treat that as approval for adjacent paths.
+
+Never delete credentials, security material, workspace database files, config files, active profiler runs, migrations, skill source, app source, conversation records, memory graph nodes or segments, `journal/`, `data/reflections/`, PKB files, backups, or backup keys unless the user explicitly names that path and accepts the consequence.
+
+## Inspection Procedure
+
+Use local/container-visible inspection first. Prefer `df -h` on the current workspace path and on `VELLUM_WORKSPACE_DIR` when that variable is available. In Docker/container mode, `/workspace` is the persistent volume and cleanup should normally focus there.
+
+Use `du` one level at a time and sort by size to identify large directories before drilling deeper. Keep each pass readable and bounded to the volume or workspace that is actually full. Avoid whole-filesystem scans unless the target volume cannot be isolated.
+
+Use `host_bash` only when the sandbox cannot see the volume that is actually full and host-level inspection is necessary. Explain why host-level inspection is needed before using it.
+
+## SQLite Diagnosis Only
+
+If `data/db/assistant.db` dominates disk usage, inspect it only through read-only `sqlite3` access. Diagnostics may use `PRAGMA` and `SELECT` queries for:
+
+- `page_size`
+- `page_count`
+- `freelist_count`
+- `dbstat` object-size breakdowns
+
+The purpose is to determine whether growth comes from specific tables/indexes or from free-page bloat. Do not edit the database manually.
+
+Never run ad hoc `DELETE`, `UPDATE`, `INSERT`, `DROP`, `REINDEX`, schema changes, `VACUUM`, `PRAGMA writable_schema`, or any other mutating SQLite command from this skill. If SQLite tables are the main culprit, tell the user this needs product-owned retention or maintenance work rather than manual database editing.
+
+## Safer Cleanup Candidates
+
+Good candidates to inspect and propose, when they are clearly nonessential and user-approved, include:
+
+- Scratch or temporary downloads created by the assistant.
+- Generated build artifacts.
+- Old logs.
+- Stale temporary directories.
+- Completed profiler runs that are no longer active.
+- Stale caches.
+- Large old diagnostic text attachments such as spindumps.
+- Package cache cleanup or Docker cache/artifact cleanup after the user approves the exact package manager or Docker action.
+
+Treat source files, persistent records, and user-authored content as protected unless the user explicitly names them for deletion and accepts the consequence.
+
+## Non-Goals
+
+ATL-450 product work is out of scope for this skill. Do not implement configurable trace or audit retention jobs, attachment retention or compression jobs, upload-time image re-encoding, first-class disk-usage UI, one-click cleanup buttons, or scheduled SQLite vacuuming. If inspection shows those are needed, report that as follow-up product work after the immediate storage cleanup path is safe.

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

@@ -105,7 +105,7 @@ Use `agent` for a single sequential leaf (throws on failure). Use `parallel`/`ma
 
 | Option    | Type                       | Effect                                                                                  |
 | --------- | -------------------------- | --------------------------------------------------------------------------------------- |
-| `schema`  | JSON Schema object literal | Forces structured output via a tool. A schema leaf runs with **no tools** (pure judge/extractor). Use a plain JSON Schema literal, not Zod. |
+| `schema`  | JSON Schema object literal | Forces structured output via a tool. A schema leaf runs with **no tools** — no `file_read`/`file_list`/`recall`/`web_search`, so it **cannot read files or recall memory** (pure judge/extractor). Pass anything it must judge **inline** in the prompt; a schema leaf told to "read these files" answers from the model's prior, not real data. Use a plain JSON Schema literal, not Zod. |
 | `label`   | string                     | Short display/diagnostic label for the leaf.                                            |
 | `profile` | string                     | Overrides the model profile. Must exist in `llm.profiles` or the leaf throws. See [Listing profiles](#listing-available-profiles). |
 | `persona` | boolean                    | `true` makes the leaf speak AS the assistant (identity + memory) — use for output meant to be in the assistant's voice. Default is anonymous — use for impartial judging/extraction. |
@@ -128,8 +128,9 @@ workflow**.
 }
 ```
 
-- **Read-only baseline** (always available, no declaration, no launch prompt):
-  `file_read`, `file_list`, `recall`, `web_search`.
+- **Read-only baseline** (available to **tool** leaves, no declaration, no launch prompt):
+  `file_read`, `file_list`, `recall`, `web_search`. **A schema leaf gets none of these**
+  (it runs as a single forced-tool-choice call) — pass it inline content, never tell it to read.
 - **`web_fetch` is NOT in the baseline** — an outbound fetch is side-effecting (its
   URL can exfiltrate read data), so a leaf that must fetch a URL has to declare
   `"web_fetch"` in `capabilities.tools`.

package/src/config/call-site-defaults.ts

@@ -69,9 +69,18 @@ export const CALL_SITE_DEFAULTS: Record<LLMCallSite, CallSiteDefaultConfig> = {
   meetConsentMonitor: { profile: "cost-optimized" },
   meetChatOpportunity: { profile: "cost-optimized" },
   inference: { profile: "cost-optimized" },
-  // The advisor consults the strongest managed profile by default; a workspace
-  // overrides this via `llm.advisorProfile` (which floats above this).
+  // The advisor consults the strongest managed profile (`frontier`, Opus),
+  // which seeding writes into `llm.advisorProfile` on boot and floats above this
+  // layer. This static fallback — used only when no `advisorProfile` resolves —
+  // stays on the always-reserved `quality-optimized` so it can never resolve to
+  // a user-owned profile that happens to be named `frontier`.
   advisor: { profile: "quality-optimized" },
+  // Vision captioning for the image-fallback plugin. No pinned profile — the
+  // plugin resolves a vision-capable profile itself via `doesSupportVision` and
+  // passes it as an `overrideProfile`, so the call-site default is a fallback
+  // that inherits the workspace default. Pinning a managed profile would break
+  // BYOK installs where managed profiles are uncredentialed.
+  vision: {},
 
   heartbeatAgent: {
     profile: "cost-optimized",

package/src/config/feature-flag-registry.json

@@ -290,14 +290,6 @@
       "description": "Show the Velvet theme option in the macOS appearance settings. Velvet is a dark-mode variant with red/pink accent colors.",
       "defaultEnabled": false
     },
-    {
-      "id": "query-complexity-routing",
-      "scope": "assistant",
-      "key": "query-complexity-routing",
-      "label": "Query Complexity Routing",
-      "description": "Automatically route user messages to the most appropriate inference profile based on query complexity. Simple queries use the speed profile, complex queries escalate to the quality profile. The user is notified of each switch and can opt out by pinning a profile on the conversation.",
-      "defaultEnabled": false
-    },
     {
       "id": "queue-steering",
       "scope": "assistant",

package/src/config/profile-dispatchability.ts

@@ -0,0 +1,11 @@
+export type DispatchableProfileEntry = {
+  provider?: unknown;
+  model?: unknown;
+  mix?: unknown;
+};
+
+export function isDispatchableProfile(
+  entry: DispatchableProfileEntry,
+): boolean {
+  return entry.provider != null || entry.model != null || entry.mix != null;
+}

package/src/config/schemas/call-site-catalog.ts

@@ -307,6 +307,13 @@ const CATALOG_RECORD: CatalogRecord = {
       "Stronger reviewer model consulted mid-task to shape or pressure-test the plan.",
     domain: "skills",
   },
+  vision: {
+    id: "vision",
+    displayName: "Vision",
+    description:
+      "Captions images via a vision-capable profile for text-only model fallback.",
+    domain: "skills",
+  },
   homeGreeting: {
     id: "homeGreeting",
     displayName: "Home Greeting",

package/src/config/schemas/llm.ts

@@ -24,6 +24,7 @@ export const LLMProvider = z
     "openai-compatible",
     "minimax",
     "atlascloud",
+    "together",
   ])
   .meta({ id: "LLMProvider" });
 type LLMProvider = z.infer<typeof LLMProvider>;
@@ -79,6 +80,7 @@ export const LLMCallSiteEnum = z.enum([
   "meetChatOpportunity",
   "inference",
   "advisor",
+  "vision",
   "trustRuleSuggestion",
   "homeGreeting",
   "homeSuggestedPrompts",

package/src/config/schemas/memory-lifecycle.ts

@@ -169,7 +169,7 @@ export const MemoryMaintenanceConfigSchema = z
       .positive("memory.maintenance.intervalMs must be a positive integer")
       .default(24 * 60 * 60 * 1000)
       .describe(
-        "Minimum interval between database maintenance (VACUUM / PRAGMA optimize) runs, in milliseconds",
+        "Minimum interval between database maintenance (PRAGMA optimize / WAL checkpoint) runs, in milliseconds",
       ),
     quietPeriodMs: z
       .number({ error: "memory.maintenance.quietPeriodMs must be a number" })
@@ -177,10 +177,12 @@ export const MemoryMaintenanceConfigSchema = z
       .nonnegative("memory.maintenance.quietPeriodMs must be non-negative")
       .default(3 * 60 * 60 * 1000)
       .describe(
-        "Database maintenance is deferred unless at least this many milliseconds have elapsed since the last user message, so the VACUUM's exclusive lock never collides with an active user (0 disables the quiet-period gate)",
+        "Database maintenance is deferred unless at least this many milliseconds have elapsed since the last user message, so maintenance's write locks never collide with an active user (0 disables the quiet-period gate)",
       ),
   })
-  .describe("Database maintenance (VACUUM / PRAGMA optimize) scheduling");
+  .describe(
+    "Database maintenance (PRAGMA optimize / WAL checkpoint) scheduling",
+  );
 
 export type MemoryJobsConfig = z.infer<typeof MemoryJobsConfigSchema>;
 export type MemoryRetentionConfig = z.infer<typeof MemoryRetentionConfigSchema>;

package/src/config/schemas/timeouts.ts

@@ -42,6 +42,30 @@ export const TimeoutConfigSchema = z
       .describe(
         "Timeout for waiting on the LLM provider's streaming response (seconds)",
       ),
+    backgroundTurnTimeoutSec: z
+      .number({ error: "timeouts.backgroundTurnTimeoutSec must be a number" })
+      .int("timeouts.backgroundTurnTimeoutSec must be an integer")
+      .positive("timeouts.backgroundTurnTimeoutSec must be a positive integer")
+      .max(
+        2147483,
+        "timeouts.backgroundTurnTimeoutSec must be at most 2147483 (setTimeout-safe limit)",
+      )
+      .default(1800)
+      .describe(
+        "Hard timeout for heartbeat and generic background agent turns (seconds)",
+      ),
+    scheduleTurnTimeoutSec: z
+      .number({ error: "timeouts.scheduleTurnTimeoutSec must be a number" })
+      .int("timeouts.scheduleTurnTimeoutSec must be an integer")
+      .positive("timeouts.scheduleTurnTimeoutSec must be a positive integer")
+      .max(
+        2147483,
+        "timeouts.scheduleTurnTimeoutSec must be at most 2147483 (setTimeout-safe limit)",
+      )
+      .default(1800)
+      .describe(
+        "Hard timeout for deliberately-launched scheduled (talk-mode) agent turns (seconds)",
+      ),
   })
   .describe("Timeout configuration for various operations");