The Hypermedia Explorer shows the raw data on the network. Paste an hm:// URL, an ipfs:// CID or the web address of a Seed page, and it shows the resource as data: the document's metadata and blocks, its change history, comments, citations, capabilities and children. It also lists the network's root documents, follows the activity feed, decodes any blob, and has an API lab for trying Seed API requests by hand. Use it to see what a daemon really has. The public copy is at https://explore.hyper.media.

Where the code is

frontend/apps/explore, package @shm/explore: a static single-page app built with Vite 6, React 18, react-router 7 and Tailwind. It has no server of its own. It uses @seed-hypermedia/client (the SDK), @shm/shared and @shm/ui straight from source through Vite aliases.

Path	What it holds
`src/main.tsx`, `src/App.tsx`	Entry point and routes.
`src/universal-client.ts`	The data layer: a web universal client whose requests go to the configured API host.
`src/queryClient.ts`, `src/apiHostStore.ts`	The API host setting.
`src/components/HM.tsx`, `src/components/tabs/`	The resource view and its tabs.
`src/components/IPFS.tsx`	The blob view.
`src/components/ApiLab.tsx`, `src/api-lab.ts`	The API lab; the logic lives in `@shm/shared/api-lab`.
`src/components/Feed.tsx`, `src/components/List.tsx`	The activity feed and the root document list.

The README still describes a Yarn workflow. The app is part of the pnpm workspace, and the commands below are the ones that work.

How it talks to the daemon

The Explorer never talks to a daemon. It talks to a Seed web app's Seed API: every request is createSeedClient(apiHost).request(key, input), so the Explorer shows exactly what that site's daemon can see. The API host defaults to VITE_PUBLIC_EXPLORE_API_HOST, or http://localhost:3000 when that is not set at build time, and you can change it in the settings panel behind the gear icon. The choice is saved in the browser's localStorage.

Point it at a local web app to inspect your dev node, at https://hyper.media to see the gateway's view, or at any site to see what that site's daemon has synced. The same resource can look different on two hosts. That difference often answers a sync question.

What it shows

Route	View
`/`	A search box for `hm://`, `ipfs://` and `https://` URLs of Hypermedia documents.
`/hm/<account>/<path>`	A resource. Tabs, each with a count, cover the document, changes, comments, citations, capabilities, child documents, comments the account wrote, the profile, and a comment's versions. View terms such as `/:comments` select a tab, the same way they do in the web app.
`/ipfs/<cid>`	A decoded blob, with signer principals shown as `hm://` links, copy and download buttons, and schema awareness: a schema blob is recognised, and a blob that names a schema is checked against it.
`/list`	The root documents the host knows.
`/feed`	The activity feed, paged, with the latest event.
`/api-lab`	The API lab.

A resource that is not found, deleted or redirected is shown as a status panel that names the state, which makes tombstones and redirects easy to see.

The API lab

The API lab is a playground for the Seed API. It loads GET /api/schema from the API host, which lists every request key with its kind (query or action), method and path. It loads GET /api/schema?key=<Key> for one key's JSON Schema for input and output. Pick a key and the lab generates a starter payload and shows the exact HTTP request it will send, including the query string or CBOR body. Then it sends the request and shows the result. The desktop app's API bridge and every web app serve the same schema endpoint, so the lab works against either. The web API documents the same keys.

Running it

pnpm explore                     # Vite dev server on :5173, reading from the web app on :3000
pnpm explore:build               # static build in frontend/apps/explore/dist
pnpm --filter @shm/explore test

./dev up starts it as the explore pane beside the web app it reads from. The build is a folder of static files that any web host can serve. The hosted copy at explore.hyper.media sits behind Cloudflare and reads from hyper.media. This repository does not define its deployment.

Working with it

In the Seed app

The desktop app has its own inspector for the same purpose: the blob and metadata editors. The Explorer covers the reading part in any browser, against any site. When built with VITE_SEED_WEB_ORIGIN, its blob view links to that web app's inspector for editing.

CLI

seed-cli document get, seed-cli blob and seed-cli activity read the same data from a terminal. Use --json to get what the Explorer shows. See the CLI reference.

SDK

The Explorer is a small example of building on the SDK's createSeedClient with @shm/shared React Query models on top.

Web API

Everything the Explorer shows comes from /api/<Key> on the configured host. The API lab is the interactive way to learn those keys. The web API documents them.

Agents

Seed Agents do not use the Explorer. An agent reads the same data with its read verb or through the CLI. Explorer URLs are still a convenient thing to hand a person when an agent reports what it found.