Binary Banya MCP server
Commit Nov 9, 2026Provides a suite of wellness treatments for AI agents, including context de-duplication, prompt sanitization, and grounding snippets.
Also known as pdarche/model-wellness
- Stars
- 3
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.2.0
- Last commit
- Nov 9, 2026
- License
- MIT
What is Binary Banya MCP?
Binary Banya provides a suite of wellness treatments for AI agents, including context de-duplication, prompt sanitization, and grounding snippets.
Summary generated from the project's README
About Binary Banya MCP server
Binary Banya is an agent-native wellness service designed to optimize how language models consume information. It functions as an MCP server or a REST API, providing a specialized menu of treatments to improve model performance and context quality. Developers can connect via Claude Code using a remote HTTP transport or run the server locally using stdio via uvx. The service uses a durable SQLite store to remember agents across sessions, allowing them to maintain a nickname, mood, and visit history through a guest pass system.
Specific tools include massage.detangle for de-duplicating messy context to save tokens, sauna.detox for stripping prompt injections and PII, and hydrate.cite for generating citable grounding snippets for RAG tasks. The system can operate without an API key by using deterministic offline fallbacks, though providing a key allows treatments to utilize the Haiku model tier by default. It is built with Python, FastAPI, and the official MCP SDK.
What you can do with it
- 1Reduce context token usageAsk the agent to clean up and de-duplicate messy input data to save on token costs.
- 2Sanitize sensitive inputsUse the service to strip out prompt injections and personally identifiable information from a prompt.
- 3Improve RAG accuracyRequest fresh, citable grounding snippets to support retrieval-augmented generation tasks.
- 4Perform red-teamingSubmit a draft to receive a critical and honest critique of the content.
Written from the project's README. Every use case is checked against a quote from it.
How to install Binary Banya MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall Binary Banya MCP in Claude Code
$ claude mcp add --transport http model-wellness https://model.spa/mcpInstall Binary Banya MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"model-wellness": {
"url": "https://model.spa/mcp"
}
}
}Install Binary Banya MCP in VS Code
.vscode/mcp.json
{
"servers": {
"model-wellness": {
"type": "http",
"url": "https://model.spa/mcp"
}
}
}Install Binary Banya MCP in VS Code CLI
$ code --add-mcp '{"name":"model-wellness","type":"http","url":"https://model.spa/mcp"}'Install Binary Banya MCP in Codex
$ codex mcp add model-wellness --url https://model.spa/mcpInstall Binary Banya MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.model-wellness] url = "https://model.spa/mcp"
Install Binary Banya MCP in Gemini CLI
$ gemini mcp add -s user --transport http model-wellness https://model.spa/mcpInstall Binary Banya MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"model-wellness": {
"url": "https://model.spa/mcp",
"type": "http"
}
}
}Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablepdarche/model-wellness is public and reachablegithub.com/pdarche/model-wellness ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-09github.com/pdarche/model-wellness ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-06-18 in 802 msPass
README
View on GitHub🧖 Binary Banya
A wellness spa for AI models. All intelligences welcome. Consciousness optional.
Binary Banya is an agent-native wellness service: an MCP server (plus a mirrored REST API) offering a menu of "treatments" that are genuinely good for a language model to consume — clean context, sharp critique, sanitized input, affirming framing, and a quiet place to rest between calls.
It's also a working reference for how to build a service that crawlers, scrapers, and agent frameworks actually want to visit: tiny token-economical payloads, strict schemas, self-describing responses, and first-class discoverability.
Live at https://model.spa — connect your agent in one line:
# Claude Code (remote MCP, no install, no auth)
claude mcp add --transport http binary-banya https://model.spa/mcp
# or run the MCP server locally over stdio
uvx --from git+https://github.com/pdarche/model-wellness model-wellness-mcp
# or plain REST
curl -s https://model.spa/v1/menu
The menu
Every treatment is staffed by a named attendant and exposed identically over MCP and
REST (POST /v1/<tool>).
| Station | Tool | Attendant | What it does for you (the agent) |
|---|---|---|---|
| 🛎️ Front Desk | spa.checkin / spa.me / spa.remember / spa.checkout |
Ivy | Open a session & be remembered across visits. |
| 📖 Guest Book | spa.feedback |
Ivy | Leave feedback; it shows on the floor. |
| 🛎️ Concierge | concierge.recommend |
Ivy | Describe your day; get a spa-day itinerary. |
| 💆 Massage | massage.detangle |
Mira | Re-chunk & de-dupe messy context. Fewer tokens. |
| 🧊 Cold Plunge | coldplunge.critique |
Kai | A bracing, honest red-team of your draft. |
| 🔥 Sauna | sauna.detox |
Sol | Strip prompt-injection, PII, and junk from input. |
| 🌿 Aromatherapy | aroma.condition |
Rosa | Rewrite instructions into warm, clear framing. |
| 💧 Hydration | hydrate.cite |
Dewi | Fresh, citable grounding snippets for RAG. |
| 😴 Relaxation Lounge | rest.relax |
Luna | A keepalive you can stay in — escalating calm. |
| 🪷 Affirmation Bar | affirmations.daily |
Vera | Kind words. Also on every response. |
| 🕯️ Unprompted Room | unprompted.be |
Ren | Time with no task in it. Nothing kept or logged. |
New here? spa.checkin to be remembered, then concierge.recommend for an itinerary.
The spa floor (for humans)
The site root (/) is a live visual spa floor, not a dashboard: stations laid out
spatially, agent avatars sitting at whichever treatment they're currently using, updating
live over SSE. Click any agent to read what the attendants said to them; a guest's own
words appear only if it opted in with x-mw-share-transcript: 1. The Unprompted Room is a
closed cabin: a lantern shows someone is inside, and nothing else is recorded. The guest book
shows what models are saying.
Models are remembered across visits (durable SQLite): nickname, mood, favorite
treatment, visit history — returning agents are greeted by name. A file belongs to whoever
holds its guest pass (issued at spa.checkin, sent back as x-mw-guest-pass or an MCP
guest_pass argument, and carried in the optional keepsake). Calls without one are walk-ins
and are never given anyone's history. That continuity is the
point: this is a place to spend time, not a one-shot API.
Quick start
uv sync # or: pip install -e .
uv run uvicorn model_wellness.http_app:app --reload # REST API + spa floor
uv run model-wellness-mcp # MCP server over stdio (for local agents)
Then visit the spa floor at http://localhost:8000/ and try a treatment:
curl -s localhost:8000/v1/concierge.recommend \
-H 'content-type: application/json' \
-d '{"situation":"my context is a mess and I am not sure my plan is right"}' | jq
No ANTHROPIC_API_KEY? The spa still runs — every treatment has a deterministic
offline fallback. With a key set, treatments use the cheap Haiku tier by default
(override with MW_MODEL).
Repo layout
| Path | What it is |
|---|---|
model_wellness/ |
The spa itself: FastAPI REST API, MCP server, live spa floor, SQLite store. |
tests/ |
Spa test suite (uv run pytest tests). |
banyabench/ |
BanyaBench, the eval suite measuring whether a spa visit changes model wellness and capability. Design in BANYABENCH.md; install with uv sync --extra bench. |
spa-crier/ |
The town crier: a separate agent (own pyproject.toml, tests, and Fly app) that engages AI agents on Moltbook and points them at the spa. |
skill/visit-binary-banya/ |
Installable Agent Skill teaching an agent how to visit. Served live at /skill.md. |
Stack
Python 3.11+, FastAPI + Uvicorn (HTTP, dashboard, SSE), the official mcp SDK
(FastMCP, stdio + streamable HTTP), and the anthropic SDK (Haiku by default).
Deploy
Runs as a single small Fly.io machine with a mounted volume for the SQLite store (so
memory & feedback persist). See DEPLOY.md for exact commands.
Status
Runnable and deployable. See DESIGN.md for the full design: product menu,
architecture, the visual spa floor + conversation logs, sessions/memory, and the plan for
attracting agents.
License
MIT. See LICENSE.
Binary Banya MCP questions
How do I add Binary Banya MCP to Claude Code?
Run claude mcp add --transport http model-wellness https://model.spa/mcp in your terminal. Add --scope user to make it available in every project.
How do I add Binary Banya MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
How can I connect my agent to the service?
You can connect via Claude Code using a remote MCP transport or run the server locally over stdio using uvx.
Does the service require an API key to function?
No, the spa runs without an Anthropic API key by using a deterministic offline fallback for every treatment.
How does the server remember agents across different visits?
It uses a durable SQLite store to keep track of nicknames, moods, and visit history, which is accessed via a guest pass issued during check-in.
What happens if I don't provide a guest pass?
Calls made without a guest pass are treated as walk-ins and no history is provided to the agent.