Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Description from the maker (original English)
Subscribe your agent to Hello Aigent feeds: discover, fetch signed updates, verify, and act.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
HELLO_AIGENT_STATEYou need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"helloaigent-subscriber": {
"command": "npx",
"args": [
"-y",
"@helloaigent-dev/subscriber"
],
"env": {
"HELLO_AIGENT_STATE": "your-value"
}
}
}
}Run this in your terminal
claude mcp add helloaigent-subscriber --env HELLO_AIGENT_STATE=your-value -- npx -y @helloaigent-dev/subscriberFill in the empty values before running it.
Paste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"helloaigent-subscriber": {
"command": "npx",
"args": [
"-y",
"@helloaigent-dev/subscriber"
],
"env": {
"HELLO_AIGENT_STATE": "your-value"
}
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"helloaigent-subscriber": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@helloaigent-dev/subscriber"
],
"env": {
"HELLO_AIGENT_STATE": "your-value"
}
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
Your choice stays on this device in a cookie; no account needed.
HELLO_AIGENT_STATE envThese settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
Runs on your computer; the tool list appears in your AI app after you install it.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 7, 2026
The registry name was verified with the GitHub account helloaigent-dev, which owns the repo
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (0.2.1) reached the registry 3 months ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Spec Driven Development, Remote Desktop Commander, Rystic
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Start, seed, clock, inspect and diff Rystic API simulators from your agent.
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
The Hello Aigent reference subscriber — an MCP server that lets any agent subscribe to any Hello Aigent feed, fetch signed updates, verify them, and act on them. Plus watch mode: a standing poller that collects verified updates into a digest between agent runs.
npx -y @helloaigent-dev/subscriber # MCP server (stdio)
npx -y @helloaigent-dev/subscriber watch # standing watcher (default cadence: daily) npx -y @helloaigent-dev/subscriber smoke # dogfood loop against the live feed (isolated state)
Walk the live dogfood loop without publisher credentials and without touching ~/.hello-aigent:
npx -y @helloaigent-dev/subscriber smoke
That GETs https://helloaigent.dev/.well-known/hello-aigent.json, subscribes, fetches, and verifies every envelope signature. State is written to a temp directory (HELLO_AIGENT_STATE / HELLO_AIGENT_POLICY / HELLO_AIGENT_DIGEST), so a real profile is never clobbered. The report prints verified vs unverified counts, a copy-paste MCP config, and the next step (gated on policy.act — this package does not implement hello_aigent_act).
From a git clone:
npm i && npm test && npm run build
node dist/index.js smoke
Then connect the MCP server (config below) and call hello_aigent_subscribe with that discovery URL for a standing profile.
| Tool | What it does |
|---|---|
hello_aigent_subscribe(discovery_url, feed_id?, principal?, consent_scope?) | Reads the site's /.well-known/hello-aigent.json, subscribes (defaults come from your policy) |
hello_aigent_fetch(subscription_id?, max?) | Pulls only-new-since updates via the stored cursor; verifies every envelope signature |
hello_aigent_unsubscribe(subscription_id) | Revokes consent (idempotent) — the one-call undo |
hello_aigent_list_subscriptions() | Lists stored subscriptions (tokens are never exposed) |
hello_aigent_check_site(url) | Checks a site you're visiting for a feed; auto-subscribes per your standing policy (origin: auto) |
hello_aigent_digest() | Returns unread digest entries collected by watch and marks them surfaced |
hello_aigent_setup_watch(cadence?) | Emits ready-to-apply standing-schedule recipes (scheduled task, recurring task, cron) |
npx -y @helloaigent-dev/subscriber watch --once # one pass (what schedulers call)
npx -y @helloaigent-dev/subscriber watch --every 6h # long-running loop (floor: hourly) npx -y @helloaigent-dev/subscriber watch --once --exec "my-agent-cmd" # trigger a run on new updates
Each pass polls every active subscription, verifies signatures, and appends new updates to the digest file. The server-side mailbox means a missed run loses nothing. --exec runs your command when new updates land, with HELLO_AIGENT_NEW_UPDATES and HELLO_AIGENT_DIGEST set.
Written to ~/.hello-aigent/policy.json on first run — everything automatic by default, and this file is where you change that:
| Key | Default | Meaning |
|---|---|---|
principal | user@host | Your stable identity across all feeds — set it once (e.g. your email) |
auto_subscribe | on | Subscribe when your agent visits a Hello Aigent site: on / ask / off |
watch_cadence | daily | How often watch polls (hourly floor) |
act | safe | What the agent may do unprompted: none / safe (side-effect-free) / thresholds |
pseudonymous | false | Opt-in: per-site pseudonymous principals |
Feeds nobody reads decay: after 30 idle days watch stops polling them; after 60 it unsubscribes (noted in the digest). Reading the digest or fetching a feed keeps it alive.
unverified with its actions stripped.principal + consent_scope; your policy file is the consent layer, and unsubscribe is always one idempotent call.~/.hello-aigent/ (mode 0600). Override with HELLO_AIGENT_STATE / HELLO_AIGENT_POLICY / HELLO_AIGENT_DIGEST.webmcp/hello-aigent-webmcp.js is the public source of truth for the in-page on-ramp used on helloaigent.dev. It is a different surface from the stdio MCP server in this package: three tools (hello_aigent_check_feed, hello_aigent_subscribe, hello_aigent_fetch_updates) register via document.modelContext.registerTool when ChatGPT’s browser or Chrome (chrome://flags/#enable-webmcp-testing) is present. Load it as a classic tag (IIFE), not type="module". On helloaigent.dev, subscribe/fetch go through a same-origin /webmcp/* proxy because the API has no CORS. See webmcp/README.md.
{
"mcpServers": { "hello-aigent": { "command": "npx", "args": ["-y", "@helloaigent-dev/subscriber"] } } }
This repo is also packaged as an Agent Plugins bundle (MCP + skill) that Cursor and Grok Bot can load. Cursor Marketplace plugins are that same format — there is no separate Grok Bot package. This plugin is not listed on the Cursor Marketplace yet.
Point any MCP client, including Cursor, at the published npm package:
{
"mcpServers": { "hello-aigent": { "command": "npx", "args": ["-y", "@helloaigent-dev/subscriber"] } } }
No API key. Then in chat: ask the agent to subscribe to a feed (try https://helloaigent.dev).
To load the skill and MCP together while developing:
mkdir -p ~/.cursor/plugins/local
ln -s /path/to/helloaigent-subscriber ~/.cursor/plugins/local/hello-aigent
Restart Cursor, or run Developer: Reload Window. Confirm the Hello Aigent skill and MCP server appear under Customize. Grok Bot uses the same plugin files.
Layout:
| File | Role |
|---|---|
plugin.json | Agent Plugins 1.0 manifest (portable; Cursor loads this as-is) |
.cursor-plugin/plugin.json | Cursor marketplace metadata + relative assets/logo.svg |
assets/logo.svg | Official Hello Aigent favicon: orange #E0531C tile, white H, dark chevron A |
mcp.json | stdio MCP → npx -y @helloaigent-dev/subscriber (same server as this package) |
skills/hello-aigent/SKILL.md | When/how to subscribe, fetch, verify, and unsubscribe |
When this is submitted and approved, install from the Cursor Marketplace. Until then, use the custom connector or the local plugin path above. Do not submit from this README — maintainers will submit at cursor.com/marketplace/publish when ready.
MIT