Skip to content
assetlib

Assetlib privacy notice

Privacy, plainly.

Assetlib is operated by Agentica, LLC, a Delaware limited liability company. Agentica, LLC (“we,” “us,” or “our”) is responsible for the personal information described in this notice. This page explains what we collect, where it is stored, and how to ask about it.

Last updated October 24, 2026. Applies to the website at www.assetlib.dev, the console at console.assetlib.dev, the Roam, Daylight, and Still sample apps, the SDKs, and the audit command and agent plugin.

The website has no accounts, forms, or analytics.

www.assetlib.dev has no account, no sign-up form, no product analytics, and no advertising trackers. Fonts are served from this site, not from a font network. We set no cookies on this domain.

Vercel hosts the site. Like any host, it may process request information such as your IP address, user agent, and the pages you request in its access logs. See the Vercel privacy policy.

The homepage embeds the Roam, Daylight, and Still sample apps in frames served from their own addresses, so loading the homepage also loads those apps. What they store is described in section 03.

Downloading the integration checklist or demo walkthrough requires no form submission. Ordinary hosting logs still apply.

The console stores your sign-in identity and your artwork.

The console at console.assetlib.dev is where workspaces, images, placements, and releases live. It is a bounded preview for non-confidential artwork.

What we collect when you sign in

  • From GitHub: your GitHub account identifier, profile name, avatar, and a verified email address. We request the read:user and user:email scopes only. We never request access to your repositories.
  • From Google, if you sign in with Google: your Google account identifier, name, profile picture, and verified email address. We request the openid, email, and profile scopes only, and no other Google data.
  • Session: a session cookie that is Secure, HttpOnly, and SameSite=Lax. Sessions expire after eight hours. The OAuth access token GitHub or Google issues is stored encrypted and used only to complete sign-in. Google’s ID token is not stored.
  • Request metadata: we use IP addresses for rate limiting on sign-in, uploads, onboarding, and delivery, in short-lived counters. Each session also stores the IP address and user agent it signed in from, which you can see under Settings › Sessions in the console.

What we store when you use a workspace

  • Images you upload, including the exact original image file (kept private to your workspace) and the prepared versions we generate from it, and Lottie animation files and short looping MP4 clips, whose uploaded bytes become public when published. Raster preparation strips embedded raster metadata such as EXIF; animation layer names and descriptions you supply can remain.
  • Placements, draft bindings, release notes, release history, and optional image descriptions.
  • Workspace membership and roles, the email addresses of people you invite, and an audit history of workspace changes.
  • If you answer it, how you heard about Assetlib, picked from a short list when you create your first workspace. It is optional, is not shown to other members, and is used only to count which channels bring new workspaces.
  • App token names, short prefixes, hashes, and when each was last used. The full token is never stored.
  • Build catalogs registered with assetlib sync: placement declarations, source references (file paths and line numbers), the app version and build number, and the repository, commit, branch, and CI run details the CLI reads from your CI environment.
  • For artwork sent from the Figma plugin, the Figma file key, node ID and name, export settings, and successful send or sync time. When the connected plugin scans linked Assetlib layers, it also reports their file, page, frame and layer names and identifiers, placement and artwork links, and scan coverage to the connected workspace. These reports do not include the whole Figma document or the contents of unrelated layers, and are not public release data.
  • Your workspace’s public SDK configuration: organization and app identifiers, the environment, a delivery URL, and public verification keys with their key IDs. It contains no credentials.
  • Usage counts sent by apps whose developers turn on SDK usage reporting, described in section 04.

What becomes public

When you publish a release to staging or production, its signed manifest and the prepared artwork it references become publicly retrievable by anyone with the URL, with credentialless cross-origin access, and may be cached by browsers and content delivery networks. That is how connected apps fetch them. Unpublished artwork is not public: workspace members can see it, and holders of an app’s draft tokens, such as the Figma plugin, can read the prepared versions of every image in that app. Original image files require workspace membership; published Lottie files and clips are served as uploaded.

Where it is stored

The console runs on Vercel. Account, session, and workspace data are stored in a managed PostgreSQL database provided by Neon in the AWS US East 1 region. Image, animation, and clip files you upload, including originals and prepared versions, are stored in Cloudflare R2 object storage. These providers process the data on our behalf: see the Vercel privacy policy, the Neon privacy policy, and the Cloudflare privacy policy. Backups may retain earlier copies. We do not sell or share this data for advertising.

The console records operational logs and the workspace audit history, and runs no product analytics on the people who use it. Webhooks and Slack integrations that a workspace owner configures send event metadata to the destinations that owner chooses.

The sample apps keep their data on your device.

Roam, Daylight, and Still are small apps we built that run in your browser or on your device. They have no accounts and no analytics. Roam and Daylight are open source; Still’s source is not published.

Roam and Daylight hold the places you save and the tasks you add in memory only, until the app reloads. If you paste a public SDK configuration into one, the app stores it locally so it can reconnect, and it caches the verified artwork it downloads. Still stores your routines, session history, and selected scene locally; its scenes are bundled with the app. Clearing the app’s data, or your browser’s site data, removes what they stored.

When Roam or Daylight checks for a release, it requests the signed manifest and images from the console’s public delivery routes. Those requests carry ordinary HTTP metadata and are subject to the hosting logs described above. Still makes no requests to the console.

Roam and Daylight also have a switch on their connection screen, off by default, that turns on the SDK usage reporting described in section 04 for the workspace you connected: placement keys, counts of resolved and displayed images, the app’s version and platform, and a random install ID the app keeps on your device, never your name, account, or a device identifier. The counts go only to the workspace whose configuration you pasted, where they are kept as described in section 06. Turning the switch off, disconnecting the app, or clearing its data stops the reports; the random ID is replaced when the data is cleared.

The SDKs send no events unless your app turns reporting on.

When your app calls refresh, the SDK requests the signed manifest and any images it needs from the delivery URL in your configuration. These delivery requests carry no device identifier and no user identifier. For each environment, the console keeps when its manifest was first and last requested and which release was served, so a workspace can see that an app is reading it; it stores no address, device, or user identifier with that record. There is no background polling for releases: those requests happen only when your code asks for them.

Usage reporting is off by default. From version 0.4.0-preview.1, the JavaScript and Expo SDKs can send usage counts if your app turns reporting on: how often each placement’s image was resolved, displayed, or replaced by its bundled fallback. Reporting includes the SDK name and version, your app’s platform, version, and build number, the environment, asset and release identifiers, variant coordinates (appearance and arm), the image source or fallback reason, and an opaque per-install ID, generated by the SDK and stored inside your app or supplied by your app. Reports carry no device model, location, or image, and the SDK adds no user identity; an install ID your app supplies must not be a user or device identifier either. The Swift and Kotlin SDKs send no usage reports. The console keeps the counts as daily totals and a one-way hash of each install ID per day; it does not store raw events or raw install IDs. Totals are eligible for cleanup after 90 UTC days and hashes after seven; deletion can lag if processing fails. If you turn reporting on, your app’s own privacy disclosures need to cover it.

Delivery requests reach the console’s public routes and are logged like any other request to Vercel. The SDK caches verified artwork on the device inside your app’s own storage, within documented limits. When delivery fails, a correctly integrated app shows compatible verified cached artwork or the image bundled in the app.

You decide what your app does with the artwork. Your own privacy notice covers your app and its users; this one covers ours.

The audit runs on your machine and reports to you.

The @assetlib/audit command reads image files to measure their dimensions, sizes, and hashes and, only if you ask, scans quoted filenames in a source folder you choose and reads commit times and changed file paths from your local git history, never commit messages or authors. It makes no network calls of its own. The first run fetches the package from the npm registry, which is governed by the npm privacy policy. It prints its report to your terminal and sends nothing to us.

The report can include your file names and source locations. Review it before sharing it anywhere.

The assetlib-audit plugin for Claude Code and Codex is a set of instructions for the agent you already use; it runs the same command. The agent, not Assetlib, handles what it reads and sends, under its own provider’s terms. We receive nothing from the plugin.

If you connect a coding agent to a workspace through Assetlib’s MCP server (the assetlib plugin, or the server added by hand), we store the name the agent registers with and its return addresses, and for each connection: your account, the workspace, whether it may change the draft, when it was made and last used, and hashes of its tokens, never the tokens themselves. The agent receives what its tools return, such as placements, image previews, release history and your workspace’s public SDK configuration, and handles that under its provider’s terms. Images it uploads are stored like any upload. You can disconnect it at any time under Profile in the console.

Release history is kept so apps can roll back.

  • Account data stays for as long as the account exists. Sessions expire after eight hours; expired session records, with their IP addresses and user agents, are cleaned up when new sessions are created.
  • Artwork and releases are retained while the preview runs. Archiving an image hides it from new use but does not delete it, because an installed app may still reference a release that includes it. Archive is reversible.
  • Usage counts from apps that turn on SDK reporting are kept as daily totals and install ID hashes. Totals are eligible for cleanup after 90 UTC days and hashes after seven; deletion can lag if processing fails.
  • Agent connections end when you disconnect them, when you leave the workspace, or after 30 days without use; ended connections remain in the workspace’s activity log. An agent registration that never connects is deleted after about a day.
  • Hosting logs are kept according to Vercel’s retention, not ours.
  • At the end of the preview, or if the service is discontinued, we may delete workspaces and their data. We will say so on this site beforehand where practical.

Self-service account deletion and full workspace export are not available yet. Requests are handled by hand, as described below.

Ask, and we will act by hand.

You can ask us to show you the data we hold about you, correct it, delete your account and workspace, or export your workspace. Write to the contact below from the email address you sign in with so we can match the request. We will confirm what was done.

Deleting a workspace removes its draft and published data from the live service. Backup copies, such as database snapshots, can keep earlier data until they are deleted, and the deletion cannot recall images already cached by browsers, content delivery networks, or installed apps that fetched them while they were public.

You can revoke Assetlib’s access at any time: for GitHub from your GitHub application settings, for Google from your Google account’s third-party connections. Revoking it does not delete your data or end Assetlib sessions that are already signed in; sign out of those under Settings › Sessions in the console, or let them expire after eight hours.

If you are in a jurisdiction that gives you additional rights over your personal data, such as the EU, the UK, or California, the contact below is how you exercise them. We do not sell personal data.

Changes are dated, and someone answers.

Assetlib is not directed to children under 16, and we do not knowingly collect their data. If you believe we have, contact us and we will remove it.

When this notice changes, the date at the top changes with it. Material changes to what the console collects will also be noted on the console’s sign-in page.

Contact Agentica, LLC with questions and privacy requests: tylerzhaodev@gmail.com. For a suspected security vulnerability, follow the private reporting steps on the security page rather than email.

Assetlib is the product name. This notice describes the service as it exists on the date above.