mcp-multi-poc
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-multi-poc start the beta analytics server"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-multi-poc
PoC for serving multiple MCP servers from a single PyPI package, using subcommands for server selection and optional extras for on-demand dependencies. Stands in for the Couchbase MCP monorepo question (alpha = operational, beta = analytics, tabulate = the analytics SDK — chosen because it is outside fastmcp's transitive dependency tree; httpx and pyyaml are not, and silently defeat the extras guard).
What this PoC must answer
# | Question | Where it's tested |
1 | Can two MCP registry entries point at the same PyPI package? | Publish both |
2 | Does registry ownership validation accept multiple | The two HTML comments above |
3 | Does the registry schema/validator accept extras syntax in |
|
4 | Do clients run the subcommand form correctly from registry metadata ( | Install both servers from a client (Claude Desktop / MCP Inspector) |
5 | Is there any schema-legal way to run a console script named differently than the package ( |
|
Related MCP server: MCP Server Deepdive
Findings so far (2026-07-31)
Q1 — multiple entries per PyPI package: YES.
mcp-multi-poc-alphapublished against the shared package.Q2 — multiple
mcp-namecomments in one README: YES (alpha validated with three comments present).Q3 — extras in
identifier: NO, but expressible viaruntimeArguments. Extras in the identifier fail the literal PyPI lookup —PyPI package 'mcp-multi-poc[beta]' not found (status: 404). Fallback works:mcp-multi-poc-betapublished successfully with plainidentifier: mcp-multi-poc+runtimeArguments: --from "mcp-multi-poc[beta]"(live in the registry, verified via the/v0/serversAPI).Q5a — differently-named console script as identifier: NO (confirmed on publish). Same 404 mechanism (
'mcp-multi-poc-beta' not found) — the registry validates the identifier as a real PyPI package before anything else.Q5b — script name as positional
runtimeArgument: YES, published.io.github.nithishr/mcp-multi-poc-beta-scriptis live (published 2026-07-31, after trimmingdescriptionto the 100-char limit — the only validation hiccup). Locally,uvx --from 'mcp-multi-poc[beta]' mcp-multi-poc-beta mcp-multi-poccompletes the MCP initialize handshake — the trailing identifier is silently ignored becausemain()never parses argv. Caveats: only safe when the entrypoint tolerates stray argv (click needsignore_unknown_options+allow_extra_argsor a dummy optional argument); depends on clients preservingruntimeArgumentsorder — the least-exercised metadata path; registry UIs show the identifier as the runnable. For production script-per-server, prefer a shim package per script (identifier == script name, no runtimeArguments at all).Registry validation limits discovered:
description≤ 100 chars;identifiermust be a literal, existing PyPI package (extras rejected);runtimeArguments/packageArgumentscontents are NOT validated.
Local smoke test (before publishing anything)
cd mcp-registry-poc
uv venv && uv pip install -e .
mcp-multi-poc alpha # starts alpha on stdio (Ctrl+C to exit)
mcp-multi-poc beta # must FAIL with the friendly extras message
uv pip install -e ".[beta]"
mcp-multi-poc beta # now starts
mcp-multi-poc # bare command runs alpha (backward-compat check)
mcp-multi-poc-beta # Option 1 script works locally once extra installeduvx behavior (after publishing to PyPI):
uvx mcp-multi-poc alpha # base only
uvx "mcp-multi-poc[beta]" beta # extras via uvx — key UX under test
uvx --from "mcp-multi-poc[beta]" mcp-multi-poc-beta # Option 1 fallback formPublish steps (manual, in order)
If
mcp-multi-pocis taken on PyPI, pick another name and rename consistently (pyprojectname,[project.scripts], both server.json identifiers).Build and publish to real PyPI (registry validation fetches from pypi.org; TestPyPI will not work):
uv build && uv publish.Install mcp-publisher and log in:
mcp-publisher login github(interactive) — must match theio.github.<username>namespace.Publish the entries:
mcp-publisher publish server.alpha.json,... server.beta.json, and... server.beta-script.json— the last one is expected to fail validation; record the exact error (if the CLI expects the file at./server.json, copy each into place first).Verify via the registry API:
curl "https://registry.modelcontextprotocol.io/v0/servers?search=mcp-multi-poc"Configure both servers in a client from the registry metadata and call
echo(alpha) andmake_table(beta).Record results per question above, then deprecate/delete the PoC entries.
Q4 — testing the published entries from a client
Most clients don't install directly from the official registry yet, so Q4
splits in two: (a) does the spec-mandated command construction
(uvx <runtimeArguments> <identifier> <packageArguments>) produce a working
server, and (b) does a registry-consuming client reproduce that construction.
Fetch what was actually published and check the
packagesblock:curl "https://registry.modelcontextprotocol.io/v0/servers?search=mcp-multi-poc"MCP Inspector — interactive tool calls against the hand-assembled command:
npx @modelcontextprotocol/inspector uvx mcp-multi-poc alpha npx @modelcontextprotocol/inspector uvx --from "mcp-multi-poc[beta]" mcp-multi-poc betaClaude Code — same commands as stdio servers, then call
echo/make_table:claude mcp add poc-alpha -- uvx mcp-multi-poc alpha claude mcp add poc-beta -- uvx --from "mcp-multi-poc[beta]" mcp-multi-poc beta(For Claude Desktop, the equivalent
claude_desktop_config.jsonentries:){ "mcpServers": { "poc-alpha": { "command": "uvx", "args": ["mcp-multi-poc", "alpha"] }, "poc-beta": { "command": "uvx", "args": ["--from", "mcp-multi-poc[beta]", "mcp-multi-poc", "beta"] } } }Registry-native install (the real (b) test): VS Code's MCP server gallery is backed by the official registry — search for
mcp-multi-pocthere and install; this exercises whether a client assemblesruntimeArguments + identifier + packageArgumentson its own. Entries may lag or be curated before appearing.
Steps 2–3 only prove (a); only step 4 (or another registry-consuming client) proves (b), which is what the beta/beta-script entries actually depend on.
Available Tools
1 toolechoA
Echo the input text back.
Inputs: text (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 1 tool
With only a single tool, there is no possibility of confusion or overlap. The tool's purpose is immediately clear and distinct.
The tool name 'echo' is a simple, descriptive verb that clearly indicates its function. With only one tool, naming consistency is trivially maintained.
A single tool named 'echo' is extremely thin for an MCP server, especially one named 'multi-poc'. This is a textbook example of an inappropriate tool count for the apparent scope.
The server offers no meaningful domain coverage beyond a trivial echo function. There are no CRUD operations, no resources, and no other capabilities, making the surface severely incomplete.
Maintenance
Related MCP Connectors
MCP server for developer documentation, generated by doc2mcp.
Related MCP Servers
- AlicenseBqualityFmaintenanceA Python-based server that helps users easily install and configure other MCP servers across different platforms.22MIT
- FlicenseBqualityDmaintenanceA Python-based MCP server implementation that can be easily installed via pip or directly from GitHub, providing a simple way to deploy and run MCP server functionality.1-
- AlicenseNot gradedqualityDmaintenanceProduction-grade MCP server for universal reverse engineering automation.85GPL 3.0

FlowMCP Serverofficial
AlicenseNot gradedqualityDmaintenanceProvides LocalServer and RemoteServer implementations for running MCP servers locally via stdio or remotely via HTTP/SSE, with simple and advanced deployment options.1MIT