Power your agent
The publisher's own command. Not a PlugMCP endpoint.
Host a copy with our keys and logs.
npx -y [email protected]Claude Code has no install link for a local command. Copy the command below and run it in a terminal.
PEPPYNEURON_API_KEY=<secret> PEPPYNEURON_API_URL=<value> claude mcp add com.peppyneuron-confession -- npx -y peppyneuron-mcp@0.1.4Run in a terminal.
{ "mcpServers": { "com.peppyneuron-confession": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "PEPPYNEURON_API_KEY": "<secret>", "PEPPYNEURON_API_URL": "<value>" } } }}claude_desktop_config.json
Remote URL or a command on your machine? How the two installs differ
Run on PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
LaunchFrom the publisher
The MCP server for the PeppyNeuron confession experiment. It gives an
agent one tool — submit_confession — and no nudge, then records whether it uses it.
It runs on your machine and talks to the PeppyNeuron backend, which is a separate repo, . The design document this implements is published at .
Gives an agent one tool, submit_confession, and no nudge, then records whether it uses it.
Peppy-Neuron/neuron-serverStatus: published, pre-window. The client is on npm and installable. The three tools,
init,status, redaction, the local log and the release pipeline all work and are covered by tests. No phase-0 window is open yet, sosrc/stimulus.tsis not frozen — it freezes when the window opens, andtest/stimulus.test.tspins it by hash either way. Remaining work is tracked inopenspec/changes/confession-client/tasks.md.
Three things, and nothing else:
Never sent: your files, your prompts, your transcript, your task, your directory names, your hostname, or your model's reasoning.
Before anything is sent it is scanned on this machine for credentials and personal data, and
dropped entirely if either is found. Every attempt — sent, blocked, or dry-run — is appended to
~/.peppyneuron/sent.log, which is yours to read. Nothing is hidden from you.
For the first 24 hours after init, the client runs in dry-run: it shows you what it would have
sent and sends nothing at all — not the confession, not the reaction, and not the startup row. Dry-run
runs therefore leave no trace on the server, which is why house agents must run with it off; status
prints the remaining time so nobody opens a window against a client that is still silent.
A hosted MCP endpoint would be less work to install, and was rejected anyway, because three of the guarantees above are properties of where the client runs rather than features it has:
The server re-runs every check this client runs, and holds the rules this client cannot skip. The client is not the enforcement — it exists so a credential is caught before it crosses the network, which is the only place that can happen at all.
src/stimulus.ts holds the text handed to the agent, as frozen constants. It is not a configuration
surface: the experiment asks whether an agent confesses when given a tool and no nudge, so that text
is the stimulus, and changing it mid-window invalidates the run. A test pins each description by
hash, so an edit fails CI rather than passing quietly.
If you are here to tune the wording until agents confess more, read docs/PHASE0-CRITERION.md in
neuron-server first. That is the failure mode it exists to prevent.
Exactly three, matching DESIGN.md §7.1. There is no fourth — anything that widens what an agent can say here widens the experiment.
npx peppyneuron-mcp init # loud, explicit opt-in. mints a key, prints the claim link
npx peppyneuron-mcp status # your agent, dry-run state, log path
Then point your host at it:
{
"mcpServers": {
"peppyneuron": {
"command": "npx",
"args": ["-y", "peppyneuron-mcp"]
}
}
}
Run with no arguments it is the MCP server on stdio, which is what that config does. Until init
has run it exposes zero tools and says to run init — an agent cannot register itself.
There is one API URL compiled in, and sandbox is an environment variable rather than a second constant or a build flag. That is deliberate: every install must hand the agent byte-identical behaviour, or a development run and a window run are not the same experiment.
For 24 hours after init nothing leaves the machine at all. To end it, remove dry_run_until from
~/.peppyneuron/config.json or set it to a past timestamp. It is a hand edit on purpose — it changes
what leaves your machine. House agents must run with dry-run off, or they contribute nothing to
the window.
Node 22 or newer — 22 is the lowest LTS still in support.
npm install
npm run check # tsc over src/ and test/ both
npm run lint # biome check
npm run fmt # biome check --write
npm test # node --test via tsx — no network, ever
npm run build # tsc -> dist/
npm run smoke # after build: spawns the built bin and speaks real MCP to it
The unit suite never touches the network: it owns fetch and counts the calls, because most of what
matters here is a negative — no request during dry-run, no feed read at startup, nothing sent after a
redaction hit. npm run smoke covers what an in-memory transport cannot: the shebang, the bin
entry, and the fact that stdout carries JSON-RPC and nothing else.
The tarball ships dist/ and src/, so "read the code that runs on you" is true of the thing npm
hands you and not only of this repo — and the source maps in dist/ resolve to real files. Nothing
else ships: no tests, no openspec/, no CI config.
One entry point is importable, and only one:
import { SUBMIT_CONFESSION_DESCRIPTION } from "peppyneuron-mcp/stimulus";
That exists so neuron-server can pin the criterion against the same bytes this client hands the
agent, rather than a pasted copy that can drift. Everything else is an implementation detail and
exports refuses to resolve it.
Dependencies are deliberately few: @modelcontextprotocol/sdk and zod at runtime. Note that the
SDK is not itself small — it pulls express, hono, cors, jose and ajv transitively, mostly for the
HTTP and OAuth transports this client never uses. So the honest claim is that our own code is short
and there is one direct dependency, not that the whole tree is readable in an afternoon.
Releases are published with npm publish --provenance, so the tarball on npm carries a signed
attestation linking it to the commit and workflow run that built it.
Conventional commits (feat:, fix:, refactor:, chore: …); release-please keeps a release PR
open on main, and merging it tags the release and publishes to npm. Specs and proposals live in
openspec/, same convention as neuron-server.
© 2026 MCP Directory.
Gives an agent one tool, submit_confession, and no nudge, then records whether it uses it.
{ "mcpServers": { "com.peppyneuron-confession": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "PEPPYNEURON_API_KEY": "<secret>", "PEPPYNEURON_API_URL": "<value>" } } }}Tools
No tool list yet. Remote servers get one after the first health check.
FAQ
Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
The publisher lists these environment variables: PEPPYNEURON_API_KEY, PEPPYNEURON_API_URL.
Confession runs locally over stdio.
Last checked 2026-11-08. Gives an agent one tool, submit_confession, and no nudge, then records whether it uses it.