@agent-native/core 0.122.4 → 0.123.1

package/corpus/templates/chat/.agents/skills/sharing/SKILL.md

@@ -0,0 +1,228 @@
+---
+name: sharing
+description: >-
+  Framework-level sharing and privacy for user-authored resources
+  (dashboards, documents, forms, decks, etc.). Use when making a resource
+  table ownable, wiring list/read/update access checks, or dropping the
+  standard share dialog into a template.
+scope: dev
+metadata:
+  internal: true
+---
+
+# Sharing — Private by Default, Explicit Share
+
+## Rule
+
+Any resource a user **creates** (dashboards, documents, forms, decks, compositions, booking links, issues, analyses) is **private to the creator** by default and visible to others only when they have been **explicitly shared** with or when the creator changes visibility to `org` or `public`.
+
+This is the framework-level primitive. Every ownable resource gets it for free — same API, same UI, same skill.
+
+## Concepts
+
+### Three visibility levels
+
+- **`private`** — owner + explicit share grants only. Default.
+- **`org`** — owner + explicit grants + anyone in the same org (read-only).
+- **`public`** — owner + explicit grants + **anyone with the link** (read-only). Public docs do NOT appear in other users' list/sidebar/search results — `accessFilter` omits them by default. They're reachable by id (`resolveAccess` admits them) so direct links and SSR routes like `/p/:id` keep working. If a list endpoint legitimately needs cross-user public discovery (a template gallery, etc.), pass `accessFilter(table, shares, ctx, minRole, { includePublic: true })`.
+
+Visibility is coarse. Explicit share grants are fine-grained (per user or per org).
+
+### Roles on a share grant
+
+- **`viewer`** — read only.
+- **`editor`** — read + write.
+- **`admin`** — read + write + manage shares. Does NOT replace the single `owner_email` on the resource.
+
+### Anonymous public URLs stay separate
+
+Form "publish" slugs, booking-link slugs, any feature that exposes a URL to unauthenticated users — these are a different axis and are NOT controlled by the sharing system. Keep them alongside it.
+
+## Make a resource ownable
+
+In your template's `server/db/schema.ts`:
+
+```ts
+import {
+  table,
+  text,
+  integer,
+  now,
+  ownableColumns,
+  createSharesTable,
+} from "@agent-native/core/db/schema";
+
+export const decks = table("decks", {
+  id: text("id").primaryKey(),
+  title: text("title").notNull(),
+  data: text("data").notNull(),
+  createdAt: text("created_at").notNull().default(now()),
+  updatedAt: text("updated_at").notNull().default(now()),
+  ...ownableColumns(), // adds owner_email, org_id, visibility
+});
+
+export const deckShares = createSharesTable("deck_shares");
+```
+
+Then register it **in `server/db/index.ts`** (not the schema file — keeps the schema file free of the `getDb` closure and avoids circular imports):
+
+```ts
+// server/db/index.ts
+import * as schema from "./schema.js";
+import { createGetDb } from "@agent-native/core/db";
+import { registerShareableResource } from "@agent-native/core/sharing";
+
+export const getDb = createGetDb(schema);
+export { schema };
+
+registerShareableResource({
+  type: "deck",
+  resourceTable: schema.decks,
+  sharesTable: schema.deckShares,
+  displayName: "Deck",
+  titleColumn: "title",
+  getResourcePath: (deck) => `/deck/${deck.id}`,
+  getDb,
+});
+```
+
+The `type` string is the stable id the UI and actions use. `getDb` is required — the framework-level share actions use it to reach your template's DB.
+
+### Restricting public visibility and cross-org user shares
+
+Some resources should NOT be reachable by an arbitrary authenticated user even with the link, and should NOT be shareable to an email outside the org. Two optional registration flags lock these axes down:
+
+```ts
+registerShareableResource({
+  type: "extension",
+  // ...
+  allowPublic: false, // hides "Public" in the share dialog and rejects it server-side
+  requireOrgMemberForUserShares: true, // user shares must target an org member or pending invitee
+});
+```
+
+- **`allowPublic: false`** — `set-resource-visibility('public')` throws `ForbiddenError`, `accessFilter` / `resolveAccess` treat any stored `'public'` row as private (defense in depth against bad data), and the share popover hides the "Public" option. `list-resource-shares` returns `policy.allowPublic: false` so the UI follows the server.
+- **`requireOrgMemberForUserShares: true`** — `share-resource` looks up `principalId` in `org_members` and `org_invitations` (pending) for the resource's `orgId` and rejects user shares to anyone else. The same flag also pins `principalType: "org"` shares to the resource's own org — sharing to a *different* org would let that org's members run code in the viewer's auth context (same threat model as a public extension). (The flag name is kept for backward compatibility; treat it as "lock both user and org shares to the resource's org".)
+
+Use both for resources that execute code or expose privileged data with the *viewer's* credentials. Extensions ship with both set: an extension's HTML calls actions / SQL / the secrets-injecting proxy as the viewer, so a public or cross-org-shared extension would let a stranger run arbitrary code with someone else's auth context. `scripts/guard-extension-no-public.mjs` (CI + `pnpm prep`) statically enforces that the extension registration keeps both flags set.
+
+Defaults match historical behaviour: `allowPublic: true`, `requireOrgMemberForUserShares: false`. Resources that don't set the flags work as before.
+
+## Filter list/read queries
+
+```ts
+import { accessFilter } from "@agent-native/core/sharing";
+
+const rows = await db
+  .select()
+  .from(schema.decks)
+  .where(accessFilter(schema.decks, schema.deckShares));
+```
+
+`accessFilter` admits rows the current user owns, has been shared on, or that the user can reach via `org` visibility. `public` rows are NOT admitted by default — see the visibility section above for why and how to opt in.
+
+## Guard write actions
+
+```ts
+import { assertAccess } from "@agent-native/core/sharing";
+
+export default defineAction({
+  schema: z.object({ id: z.string(), title: z.string() }),
+  run: async (args) => {
+    await assertAccess("deck", args.id, "editor");
+    // ...proceed
+  },
+});
+```
+
+For delete actions use `"admin"` (or fold in `"owner"` to require the real owner).
+
+## Create actions must set owner
+
+When inserting a new row, fill `ownerEmail` and `orgId` from the request context:
+
+```ts
+import {
+  getRequestUserEmail,
+  getRequestOrgId,
+} from "@agent-native/core/server/request-context";
+
+const ownerEmail = getRequestUserEmail();
+// Never fall back to a sentinel like "local@localhost" — that pools every
+// unauthenticated write into one shared tenant (see the 2026-04-29 leak and
+// guard-no-localhost-fallback). Throw / 401 when there is no session instead.
+if (!ownerEmail) throw new Error("Not authenticated");
+
+await db.insert(schema.decks).values({
+  id: nanoid(),
+  title,
+  data,
+  ownerEmail,
+  orgId: getRequestOrgId(),
+  // visibility defaults to 'private'
+  // ...
+});
+```
+
+## Drop in the share UI
+
+```tsx
+import { ShareButton } from "@agent-native/core/client/sharing";
+
+// In the resource's header/toolbar:
+<ShareButton
+  resourceType="deck"
+  resourceId={deck.id}
+  resourceTitle={deck.title}
+/>;
+```
+
+For list views, show `<VisibilityBadge visibility={row.visibility} />` next to each resource.
+
+## Actions available everywhere
+
+The framework auto-mounts these actions in every template — no per-template boilerplate:
+
+| Action                     | Args                                                                           | Purpose                                   |
+| -------------------------- | ------------------------------------------------------------------------------ | ----------------------------------------- |
+| `share-resource`           | `resourceType, resourceId, principalType, principalId, role, notify?, resourceUrl?` | Grant a user or org access. `notify` defaults to true for individual user shares; `resourceUrl` can provide the direct app link used in the notification email. |
+| `unshare-resource`         | `resourceType, resourceId, principalType, principalId`                         | Revoke access.                            |
+| `list-resource-shares`     | `resourceType, resourceId`                                                     | Current visibility + all share grants.    |
+| `set-resource-visibility`  | `resourceType, resourceId, visibility`                                         | Change to `private` / `org` / `public`.  |
+
+Both the agent and the UI use these same actions. The agent calls them as tools;
+UI code should use `ShareButton` / `ShareDialog` or the action client hooks
+instead of hand-writing route calls.
+
+## Migration pattern for existing tables
+
+When retrofitting an existing resource table:
+
+1. Add `owner_email`, `org_id`, `visibility` columns (defaults `'local@localhost'`, `NULL`, `'private'`).
+2. Backfill `owner_email` from any prior creator trail; otherwise leave the default.
+3. Add the companion `{type}_shares` table.
+4. Register via `registerShareableResource`.
+5. Update list/read actions to use `accessFilter`.
+6. Update update/delete actions to `assertAccess` with the correct role.
+7. Add `<ShareButton>` to the resource header.
+8. Add `getResourcePath` in the registration so agent-triggered shares can email a direct link even when no UI supplied `resourceUrl`.
+
+## Templates that opt out
+
+Sharing doesn't apply to:
+
+- **Personal-data apps** (mail, macros) — user-scoped by design.
+- **External source-of-truth apps** (issues → Jira, recruiting → Greenhouse) — ACL lives in the upstream system.
+- **Demo/boilerplate** (starter) — no resources.
+
+For these, add a short note to the template's `AGENTS.md` explaining why.
+
+## Analytics (follow-up)
+
+Dashboards and analyses in the `analytics` template currently live in the settings KV store (`u:<email>:dashboard-*` keys), not SQL. Sharing requires either migrating them to SQL tables (then applying this skill) or extending the settings store with a parallel share overlay. This is a tracked follow-up — see the analytics template's `AGENTS.md`.
+
+## Debugging
+
+- `ForbiddenError` from an action means the current user isn't owner / hasn't been shared / can't meet the role bar.
+- If the agent can't see a resource it just created, check that the insert actually set `owner_email` from the request context.
+- If a share doesn't take effect in the UI, confirm the template's `list-*` action uses `accessFilter` — the share rows are there but nothing is reading them yet.

package/corpus/templates/chat/.agents/skills/storing-data/SKILL.md

@@ -0,0 +1,178 @@
+---
+name: storing-data
+description: >-
+  How to store application data in agent-native apps. All data lives in SQL.
+  Use when adding data models, deciding where to store data, or reading/writing
+  application data.
+scope: dev
+metadata:
+  internal: true
+---
+
+# Storing Data — SQL is the Source of Truth
+
+## Rule
+
+All application data lives in **SQL** (SQLite locally, persistent database in production). The agent and UI share the same database. SQL stores structured records, metadata, references, and searchable text — not large raw file payloads. Do not store durable app data in the filesystem unless the app is explicitly running a Local File Mode artifact flow described below.
+
+Large binary or file-like payloads (images, video/audio, PDFs, ZIPs, screenshots, session replay chunks, thumbnails, generated assets, `data:` URLs, and base64 file bodies) must go through configured file/blob storage such as `uploadFile()` or `putPrivateBlob()`. Persist only the returned URL, asset id, or opaque blob handle in SQL. If storage is unavailable in hosted or persistent-database mode, fail closed with setup guidance instead of falling back to base64 in `application_state`, `settings`, `resources`, or app tables. Local SQLite-only dev fallbacks may exist for tiny assets, but they must be capped, documented as dev-only, and kept off hot list/read paths.
+
+**Local File Mode exception:** some artifact apps (Content, Plans, Slides, Dashboards, Designs, etc.) can intentionally use repo files as the source of truth for the artifact itself. This must be explicit via `agent-native.json`, `AGENT_NATIVE_MODE=local-files`, or an app-owned local-file action helper. In that mode, the UI and agent still go through app actions, but those actions read/write scoped files through `@agent-native/core/local-artifacts` instead of SQL rows. App state, auth, settings, credentials, collaboration metadata, and hosted database mode remain SQL. File-to-database or file-to-provider synchronization is an explicit sync step, not an implicit side effect of editing.
+
+When you add a data model, a list, or a read path, also follow the `performance` skill: project only the columns a list renders, index the columns hot queries filter/sort on, and avoid query waterfalls — so apps stay fast as data grows.
+
+## How It Works
+
+Agent-native apps use Drizzle ORM over the configured SQL backend. Local development works out of the box with a SQLite file at `data/app.db`; production and shared preview deploys need a persistent `DATABASE_URL` because container/serverless filesystems can reset. The code should behave the same across backends, but the local SQLite file is not durable once deployed.
+
+For app code, use Drizzle's schema/query DSL by default. Raw SQL is an escape hatch for additive migrations, health checks, or one-off maintenance, not the normal way to build features.
+
+### Naming migrations
+
+When you add an entry to a `runMigrations([...])` list (`@agent-native/core/db`), always give it a unique `name:` slug (e.g. `name: "analytics-alert-rules-table"`) alongside its `version`. Never renumber or reuse version numbers on existing entries.
+
+Why: version numbers alone are not a safe identity. Two branches that each independently extend the same migration list can ship different DDL under the same version numbers — whichever branch deploys first "claims" those version numbers in the bookkeeping table, and the other branch's DDL is silently treated as already applied even though it never ran. This exact collision took down analytics: parallel branches both extended their migration list through v75-v83 with different DDL, so `analytics_alert_rules`, `analytics_alert_incidents`, and `session_recordings.network_error_count` never made it to production despite the bookkeeping table showing every version as applied. A `name:` slug is tracked independently of version numbers, so it applies exactly once per database regardless of what any other branch already recorded.
+
+Existing unnamed migrations don't need to be renamed retroactively (the two gating strategies coexist), but any new entry should always carry a name.
+
+### Core SQL Stores (auto-created, available in all templates)
+
+| Store               | Purpose                                              | Access                                     |
+| ------------------- | ---------------------------------------------------- | ------------------------------------------ |
+| `application_state` | Ephemeral UI state (compose windows, navigation)     | `readAppState()` / `writeAppState()`       |
+| `settings`          | Persistent KV config (preferences, app settings)     | `getSetting()` / `putSetting()`            |
+| `oauth_tokens`      | OAuth credentials                                    | `@agent-native/core/oauth-tokens`          |
+| `sessions`          | Auth sessions                                        | `@agent-native/core/server`               |
+
+### Domain Data (per-template)
+
+Define schema with the framework Drizzle helpers in `server/db/schema.ts`. Get a database instance with `const db = getDb()` from `server/db/index.ts`. All queries are async.
+
+```ts
+import { eq } from "drizzle-orm";
+import { table, text, integer, now } from "@agent-native/core/db/schema";
+
+export const tasks = table("tasks", {
+  id: text("id").primaryKey(),
+  title: text("title").notNull(),
+  completed: integer("completed", { mode: "boolean" })
+    .notNull()
+    .default(false),
+  createdAt: text("created_at").notNull().default(now()),
+});
+
+const rows = await db.select().from(tasks).where(eq(tasks.id, taskId));
+```
+
+Never import `sqliteTable` / `pgTable` or column helpers from `drizzle-orm/sqlite-core` or `drizzle-orm/pg-core` in app templates. Use `@agent-native/core/db/schema` so the same schema can run against SQLite, Postgres, libSQL/Turso, D1, and other supported backends.
+
+| Template     | Tables                                        |
+| ------------ | --------------------------------------------- |
+| **Mail**     | emails, labels (+ Gmail API when connected)   |
+| **Calendar** | events, bookings                              |
+| **Forms**    | forms, responses                              |
+| **Content**  | documents                                     |
+| **Slides**   | decks (JSON stored in SQL)                    |
+| **Videos**   | compositions in registry + localStorage       |
+
+### Agent Access
+
+The agent uses app-specific actions to read/write the database. Core DB scripts are for inspection and maintenance, not for implementing normal product behavior:
+
+- `pnpm action db-schema` — Show all tables, columns, types
+- `pnpm action db-query --sql "SELECT * FROM forms"` — Run SELECT queries
+- `pnpm action db-exec --sql "UPDATE ..."` — Last-resort ad-hoc maintenance for short columns, multi-column writes, or computed updates when no domain action exists. For several related writes, prefer `--statements '[{"sql":"...","args":[...]}]'` so they run sequentially in one transaction. Schema changes are blocked; use reviewed additive migrations/startup code instead.
+- `pnpm action db-patch --table <t> --column <c> --where "<clause>" --find "<old>" --replace "<new>"` — **Surgical search/replace on a large text column.** Sends the diff instead of re-transmitting the whole value, so it's dramatically more token-efficient than `db-exec UPDATE` when editing multi-kilobyte documents, slide HTML, dashboard/form JSON, etc. Targets exactly one row per call — narrow `--where` by primary key. Supports `--edits '[{find,replace},...]'` for batch edits and `--all` to replace every occurrence.
+- App-specific actions for domain operations — **always prefer these over raw SQL when one exists.** They encode business rules, power the client action hooks, and for editor-backed tables (documents, slides) also push live Yjs updates to open collaborative editors. `db-patch` is the generic fallback for tables without a dedicated edit action.
+
+**For one-off maintenance, how to choose between `db-exec UPDATE` and `db-patch`:**
+
+| Scenario                                                       | Use          |
+| -------------------------------------------------------------- | ------------ |
+| `SET status = 'published'` on one row                          | `db-exec`    |
+| `SET calories = calories + 50`                                 | `db-exec`    |
+| Updating several columns at once                               | `db-exec`    |
+| Inserting/updating several rows as one logical operation        | `db-exec --statements` |
+| Fixing a typo in a 50KB markdown document's `content` column   | `db-patch`   |
+| Changing a single key in a dashboard's JSON blob               | `db-patch`   |
+| Tweaking one paragraph of slide HTML stored in `decks.data`    | `db-patch`   |
+| Any edit where you'd otherwise re-send thousands of characters | `db-patch`   |
+
+All of these honor the per-user / per-org data scoping — you can't read or write rows outside the current user's data, regardless of which tool you choose.
+
+### Frontend Access
+
+The frontend calls actions using React Query hooks from the client API. The framework owns the HTTP transport behind these hooks, so components should not call action routes with raw `fetch`.
+
+```ts
+import { useActionQuery, useActionMutation } from "@agent-native/core/client/hooks";
+
+// Read data
+const { data } = useActionQuery("list-meals", { date: "2025-01-01" });
+
+// Write data
+const { mutate } = useActionMutation("log-meal");
+```
+
+Actions are the **preferred way** for the frontend to access data. You rarely need custom `/api/` routes — only for file uploads, streaming, webhooks, or OAuth callbacks.
+
+### Production / Cloud Deployment
+
+Local SQLite works out of the box for development. To deploy to production or any environment where data must survive restarts:
+
+1. Set `DATABASE_URL` to a persistent SQL database.
+2. Set `DATABASE_AUTH_TOKEN` only when the provider requires a separate token, such as Turso/libSQL.
+3. No code changes should be needed when the schema and queries stay portable.
+
+Turso is one valid option, not the required option. Common choices include Neon or Supabase Postgres, Turso/libSQL, plain Postgres, durable SQLite, Cloudflare D1 bindings, and managed platform SQL environments when available.
+
+### Real-time Sync
+
+Polling streams database changes to the UI. When the agent writes to the database via scripts, the UI updates automatically via `useDbSync()` which invalidates React Query caches.
+
+## Do
+
+- Use Drizzle ORM for structured domain data (forms, bookings, documents)
+- Use Drizzle query builder methods (`select`, `insert`, `update`, `delete`) and portable operators from `drizzle-orm` (`eq`, `and`, `or`, `inArray`, `desc`, etc.) for app reads/writes
+- Use framework schema helpers from `@agent-native/core/db/schema`, not dialect-specific Drizzle imports
+- Use the `settings` store for app configuration and user preferences
+- Use `application-state` for ephemeral UI state that the agent and UI share
+- Use `oauth-tokens` for OAuth credentials
+- Use `uploadFile()` or `putPrivateBlob()` for large files/blob data and store only URLs, ids, or handles in SQL
+- Use core DB scripts (`db-schema`, `db-query`, `db-exec`, `db-patch`) for ad-hoc database operations
+- Use `db-exec --statements` instead of several separate `db-exec` calls for related writes; it is faster and rolls back the whole batch if one statement fails
+- Reach for `db-patch` instead of `db-exec UPDATE` whenever you're making a small change to a large text/JSON column — it's much cheaper on tokens
+
+## Don't
+
+- Don't store structured app data as JSON files
+- Don't store app state in localStorage, sessionStorage, or cookies (except for UI-only preferences like sidebar width)
+- Don't keep state only in memory (server variables, global stores)
+- Don't use Redis or any external state store for app data
+- Don't store large files, base64 blobs, `data:` URLs, screenshots, videos, audio, PDFs, ZIPs, or session replay chunks directly in SQL rows, `application_state`, `settings`, or `resources`
+- Don't implement product features with raw SQL or `getDbExec()` when Drizzle can express the query
+- Don't write SQLite-only or Postgres-only SQL in app code
+- Don't interpolate user input directly into SQL queries — use Drizzle ORM's query builder
+
+## Security
+
+- **SQL injection** — Use Drizzle ORM's query builder, never raw string interpolation for SQL queries
+- **Validate before writing** — Check data shape before writing, especially for user-submitted data
+
+## Application State and Context Awareness
+
+When storing app-state, include **navigation state** — the agent needs to know what the user is looking at. The `application_state` table holds ephemeral UI state that both the agent and UI share. Key patterns:
+
+- **`navigation` key** — the UI writes current view and selection on every route change. The agent reads this before acting.
+- **`navigate` key** — the agent writes one-shot commands to navigate the UI. The UI processes and deletes them.
+- **Domain-specific keys** (e.g., `compose-{id}`) — bidirectional state for features like email drafts.
+
+When adding a new data model or feature, also consider what navigation and selection state needs to be exposed via application-state. See the **context-awareness** skill for the full pattern.
+
+## Related Skills
+
+- **context-awareness** — How to expose navigation and selection state via application-state
+- **real-time-sync** — Set up polling so the UI updates when the database changes
+- **actions** — Create actions with `defineAction` to query the database
+- **client-methods** — Keep route details behind named client helpers/hooks
+- **self-modifying-code** — The agent can also modify the app's source code

package/corpus/templates/chat/AGENTS.md

@@ -13,7 +13,15 @@ workflow needs durable UI around the conversation.
 - Never hardcode API keys, tokens, webhook URLs, signing secrets, private Builder/internal data, customer data, or credential-looking literals. Use secrets/OAuth/runtime configuration and obvious placeholders in examples.
 - Follow the root framework contract: data in SQL, actions first, application
   state for navigation/selection, and shared agent chat for AI work.
+- Scale effort to the task. A small, well-specified change is a short read, the
+  edit, and the app's existing checks (`pnpm typecheck`, formatter, existing
+  tests) — not a codebase survey, unrequested tests, or browser automation.
 - Use actions for app operations and keep frontend/API parity.
+- Do not add `/api/*` routes for app data. If you are about to create a file
+  under `server/routes/api/`, or middleware to guard one, stop and write a
+  `defineAction` instead. The only exceptions are uploads, streaming, inbound
+  webhooks, OAuth callbacks, public unauthenticated URLs, and non-JSON
+  responses — not auth, settings, search, or CRUD.
 - Treat the chat as the default UI. When the user asks for a capability, prefer
   adding or improving the action surface first, then add a page, table, form, or
   widget only when the user needs to inspect, compare, approve, or share durable

package/corpus/templates/chat/app/entry.client.tsx

@@ -2,9 +2,12 @@
 // This keeps the ~650-700 KB gzip chat stack off the static import closure
 // of the client entry point so it can't block the first page parse.
 import { appBasePath } from "@agent-native/core/client/api-path";
+import { installRouteChunkRecovery } from "@agent-native/core/client/route-chunk-recovery";
 import { hydrateRoot } from "react-dom/client";
 import { HydratedRouter } from "react-router/dom";
 
+installRouteChunkRecovery();
+
 const basePath = appBasePath();
 const pathname = window.location.pathname;
 const routerBasePath =

package/corpus/templates/clips/.agents/skills/actions/SKILL.md

@@ -18,6 +18,8 @@ Actions in `actions/` are the **single source of truth** for app operations. The
 
 Before creating any custom REST/API route for app data, inspect `actions/` and the action table in `AGENTS.md`. If an action already exists, call it directly from the agent or with `useActionQuery` / `useActionMutation` from the UI. If the capability is missing, create or update a `defineAction`. Do not add `/api/*`, `server/routes/*`, or other pass-through endpoints whose main job is to call, repackage, or re-export an action.
 
+**Stop trigger:** the moment you are about to create a file under `server/routes/api/` — or add server middleware to guard one — stop and check it against the exception list in *When You Still Need Custom `/api/` Routes* below. If it is not on that list, write a `defineAction` instead. This applies even when you already started the route; abandoning a half-written action to hand-roll routes is the exact failure this rule exists to prevent.
+
 ## Why
 
 Actions give the agent callable tools with structured input/output, AND they give the frontend a typed client contract through hooks. One implementation serves both the agent and the UI. They keep the agent's chat context clean, they're reusable, and they can be tested independently.
@@ -390,12 +392,28 @@ This is the canonical approach for new apps. Action names must be lowercase with
 
 ## When You Still Need Custom `/api/` Routes
 
-Most operations should be actions. You only need custom routes in `server/routes/api/` for:
+This is the complete exception list. A route in `server/routes/api/` is only
+justified when the caller is not your own UI or agent, or when the payload is
+not JSON:
 
 - **File uploads** — actions receive JSON params, not multipart form data
 - **Streaming responses** — SSE or chunked responses that need direct H3 control
 - **Webhooks** — external services POST to a specific URL
 - **OAuth callbacks** — redirect-based flows that need specific URL patterns
+- **Public unauthenticated endpoints** — SEO/OG images, share links, and other
+  URLs opened directly by a browser or crawler
+- **Binary or non-JSON responses** — media, exports, generated assets
+
+Everything else is an action: CRUD, settings, search, list/detail reads, auth
+state, and anything the UI fetches as JSON. If a route needs new middleware to
+scope it to the current user, that is a signal it should be an action — actions
+already run inside request context with access checks.
+
+Existing first-party templates still contain a shrinking, explicitly
+grandfathered set of older `/api/*` CRUD routes for app data. The repository's
+`guard:no-action-twin-routes` CI check ratchets that baseline down across the
+first-party template route trees (with the separately owned Plan template
+fenced out); it is not a substitute for keeping generated apps action-first.
 
 When the agent needs a durable image or file URL, call the core `upload-image`
 action or use `uploadFile()` in server code. Do not write base64 into SQL,

package/corpus/templates/clips/.agents/skills/ai-video-tools/SKILL.md

@@ -136,6 +136,16 @@ The transcript included by `view-screen` is only a bounded preview. When its
 show where transcription ended. Call `get-recording-player-data` before
 judging completeness or quoting the full transcript.
 
+Desktop native transcripts merge the microphone and system-audio streams while
+removing overlapping duplicate speech and low-speech Whisper hallucinations. Keep
+system audio enabled when the meeting audio comes from another app.
+
+Native transcript first is a hard rule, not a preference: cloud transcription is
+fallback-only for Clips recordings. Cleanup and transcript-backed title/summary
+generation run in the durable post-finalize path, so never hide a usable native
+transcript behind failed metadata work, and keep heuristic titles replaceable
+until the agent refinement lands.
+
 Clips never routes recording/meeting audio to OpenAI for transcription. (Groq's endpoint is OpenAI-_compatible_ in request shape only — the audio goes to Groq, not OpenAI.)
 
 If no native transcript exists and no cloud fallback is available (no Builder connection and no Groq key), the action writes `status="failed"` so the UI can show a friendly prompt.
@@ -171,6 +181,38 @@ registerRequiredSecret({
 
 It is not marked `required: true` — videos still upload and play without a Groq key when video storage is connected, since native transcription (and Builder when connected) already cover the common case. The API Keys & Connections panel surfaces it so the user can add the Groq fallback if they want it.
 
+## AI setup, providers, and credits
+
+AI setup must be visible and paid-account-backed. Lead with Builder.io Connect
+for managed credits, object storage, uploads, and transcription. BYOK belongs in
+the agent sidebar's API Keys & Connections panel; template settings may signpost
+that panel but must not create a second credential vault.
+
+Which provider powers what:
+
+| Surface                                    | Provider                                  |
+| ------------------------------------------ | ----------------------------------------- |
+| Agent chat                                 | Anthropic / OpenAI                        |
+| Cleanup, titles, meeting notes             | Gemini (Builder-managed or BYOK)          |
+| Recording transcripts                      | Native first, then Builder Gemini or Groq |
+| Desktop voice dictation                    | Optional third-party speech provider      |
+
+Any optional third-party speech provider is limited to desktop voice dictation
+and is never used for recording transcripts.
+
+Use `get-builder-credit-status` when the user asks whether Builder.io credit
+limits are pausing backup transcription, transcript cleanup, summaries, or AI
+title generation. Treat an exhausted status as an FYI and upgrade path, not an
+app error: say so clearly and point the user at Builder.io credits/upgrade.
+Native browser/macOS capture remains the first transcript source regardless.
+
+## Bounded voice context
+
+Dictation cleanup, clip title/cleanup, and meeting summaries should pass bounded
+`voiceContext` to the shared cleanup/transcription path whenever active app
+context, learned vocabulary, user notes, or AGENTS.md preferences are available.
+Pass a bounded summary — never a whole transcript, document, or catalog.
+
 ## Live transcription during recording
 
 The `useLiveTranscription` hook (from `@agent-native/core/client/transcription`) runs the browser's Web Speech API alongside any recording. It accumulates final transcript text in real time with no API key required. When the user stops, the client calls `save-browser-transcript` to persist the result. `request-transcript` preserves that native result and only falls back to cloud transcription when native text is missing.

package/corpus/templates/clips/.agents/skills/brain-export/SKILL.md

@@ -0,0 +1,59 @@
+---
+name: brain-export
+description: >-
+  How `export-to-brain` sends ready Clips transcripts to Brain — single
+  exports, bounded historical imports, cursor pagination, required secrets,
+  and the retry sweep. Use when exporting transcripts to Brain, backfilling
+  past recordings, or debugging a missing or stuck Brain export.
+---
+
+# Brain Export
+
+## Rule
+
+Transcripts reach Brain through the `export-to-brain` action and the durable
+post-finalize worker — never through an ad hoc fetch to the Brain ingest URL.
+
+## Single export
+
+Use `export-to-brain --recordingId=<id>` for one ready transcript.
+
+## Bounded historical import
+
+For a bounded historical import, omit `recordingId` and pass `lookbackDays`,
+`limit`, and `concurrency`. When `nextCursor` is non-null, pass it as `cursor`
+on the next call until `nextCursor` is null.
+
+The cursor keeps the original lookback snapshot and advances by recording
+creation time plus id, so pages cannot reselect the same recordings or expand
+forever as new clips arrive.
+
+The action only selects current-user recordings with ready transcripts in the
+active organization, and reports exported, quarantined, skipped, and failed
+counts.
+
+## Required secrets
+
+Both `BRAIN_INGEST_URL` and `BRAIN_INGEST_TOKEN` must be available as scoped
+Clips secrets. Never hardcode either value — read them through the app secret
+surface.
+
+## Durable delivery and the retry sweep
+
+Transcript completion persists a pending export before handing it to the durable
+post-finalize worker. Delivery receipts include the Brain capture or sensitivity
+receipt id, while transient failures are swept and retried.
+
+The sweep also discovers ready transcripts from the last seven days that predate
+export-state tracking, in bounded batches, so recent recordings are backfilled
+after the connection is configured.
+
+Netlify builds emit a protected per-minute scheduled sweep because in-process
+intervals are not durable there. Other serverless hosts must invoke
+`runBrainExportSweepOnce` from their own scheduler.
+
+## Related skills
+
+- `ai-video-tools` — how a transcript becomes ready in the first place.
+- `recording` — the finalize path that triggers the post-finalize worker.
+- `security` — scoped secret storage for the ingest URL and token.

package/corpus/templates/clips/.agents/skills/bug-reports/SKILL.md

@@ -0,0 +1,37 @@
+---
+name: bug-reports
+description: >-
+  The embedded Clips bug-report flow — the `/bug-report` iframe launcher,
+  `/record?intent=bug-report` capture, `save-bug-report-context`, and its
+  intake limits. Use when embedding bug capture in another product, wiring the
+  launcher, or asked to collect customer bug recordings.
+---
+
+# Embedded Bug Reports
+
+## Rule
+
+`/bug-report` is an iframe-friendly launcher only. The actual capture runs
+top-level at `/record?intent=bug-report`, because browser media capture needs a
+top-level user gesture.
+
+## How it works
+
+1. The host page embeds `/bug-report` in an iframe.
+2. The launcher stores redacted host metadata through `save-bug-report-context`.
+3. Capture opens top-level at `/record?intent=bug-report`.
+4. The recording remains the canonical resource and defaults to workspace
+   (organization) visibility.
+
+## Intake limit
+
+Do not present this as anonymous customer intake until a signed intake/upload
+token flow exists — the current upload endpoints are owner-scoped, so an
+anonymous reporter cannot upload.
+
+## Related skills
+
+- `recording` — the capture and upload path behind `/record`.
+- `video-sharing` — why bug-report recordings default to organization
+  visibility instead of public.
+- `security` — redaction rules for host metadata and diagnostics.