Likerts

io.github.crosstabsPublished by GitHub account crosstabsv0.1.1Updated Nov 2, 2026

Manage open-source in-product surveys, collections, responses and exports with scoped access.

VerifiedStreamable HTTPWeb executableData & AnalyticsBusiness & Commerce

Overview

AI-generated overview

Lets an assistant manage Likerts in-product surveys, collections, responses and exports through the platform's scoped API.

What you can do
  • Create, publish and list surveys in a Likerts workspace
  • Set up collections and control whether collection access is open or closed
  • Submit, retry, read and delete survey responses
  • Export survey data
  • Use the same typed operations and authorization boundaries as the Likerts HTTP API and CLI
When to use it
Useful when you run the free, MIT-licensed Likerts survey platform and want an assistant to create or publish surveys, manage collections, or handle responses and exports without leaving the chat client.
Requirements
Either the hosted remote endpoint (workspace ID in the path) or the local npm package @likerts/mcp run over stdio with Node.js 22 or newer. Requires LIKERTS_API_URL and a scoped LIKERTS_TOKEN, sent as an Authorization bearer credential; LIKERTS_COLLECTION_TOKEN is optional for collection configuration and response submission. The local API defaults to
Before you install
The server holds workspace credentials: LIKERTS_TOKEN is a required secret and LIKERTS_COLLECTION_TOKEN is a separate collection credential, so grant only the scopes a workflow needs and keep management credentials out of embedded apps and tool arguments. Tools can create and publish surveys, submit and delete responses, and close collection access. The hosted endpoint is described as a preview for synthetic, non-sensitive evaluation data, so do not treat it as suitable for real workspace data.

Installation

In SourceWeft

  1. Open Likerts in the dashboard and add it to a workspace.
  2. Connect Authorization token credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Other MCP clients

Follow the launch instructions in the repository.

README

Likerts MCP server

Operate the free, MIT-licensed Likerts survey platform from an MCP client. The server exposes the same typed operations and authorization boundaries as the HTTP API and CLI.

The hosted server is listed in the official MCP registry as io.github.crosstabs/likerts. Configure your workspace ID and scoped service credential in your MCP client. The listing does not make workspace data public.

Install

Install a published version from npm:

sh
npm install --global @likerts/[email protected]

Alternatively, download the MCP tarball from the community release and run npm install --global /absolute/path/to/the-mcp-package.tgz. Node.js 22 or newer is required.

Configure your MCP client to run likerts-mcp over stdio. Set LIKERTS_API_URL to your API origin and supply a scoped LIKERTS_TOKEN through your environment or secret manager. The local API defaults to http://127.0.0.1:8080. Collection operations use LIKERTS_COLLECTION_TOKEN separately; keep management credentials out of embedded apps and tool arguments.

Copy the local Codex configuration or local Claude Code configuration. Both read the service credential from your environment; start with only the scopes your workflow needs.

For a source checkout:

sh
npm ci --prefix tools/mcpnpm run build --prefix tools/mcpnode tools/mcp/dist/main.js

Launch the command from an MCP client; stdout is reserved for protocol messages. Authentication and permission checks remain enforced by your API.

Remote connection

The optional hosted endpoint is https://likerts-mcp.onrender.com/mcp/{workspaceId}. It requires the workspace's scoped service credential. It is a hosted reference preview for synthetic, non-sensitive evaluation data, with the same operational and support limits as the web preview. Self-hosters use their own remote host.

Setup for Codex and Claude · Quickstart · API reference

Registry metadata and release verification

server.json matches published registry listing 0.1.1, adding npm package @likerts/[email protected] alongside the remote endpoint. The publication workflow passed on 13 September 2026 at source fe7588fabf2ee596a963ed943764021084f7d49e; independent readback matched both npm and remote metadata. Listing and npm package versions are independent. Registry 0.1.0 and the npm tarball remain immutable.

From a source checkout with Docker, Rust stable and Node 22+, run:

sh
npm ci --prefix tools/mcpnode scripts/check-mcp-registry.mjsbash -c 'source scripts/dev-env.sh; node scripts/check-mcp-lifecycle.mjs'

The registry check validates the official pinned schema, exact npm package identity and version, and published mcpName. The lifecycle check installs the public npm artifact in a temporary consumer, starts the real Rust API on loopback with disposable PostgreSQL and a restricted runtime role, discovers every tool, and uses a scoped service credential to create/publish a survey, create a collection, submit/retry/read/delete a response and close collection access. It also proves insufficient-scope denial and credential revocation. It deletes its temporary workspace/files. This does not test a specific Codex/Claude UI, backup/restore behavior or the hosted service.

Maintainers publish metadata from an exact main commit only after both required jobs in its Verify foundations push run succeed:

sh
gh workflow run publish-mcp.yml --ref main \  -f source_sha=FULL_TESTED_MAIN_COMMIT \  -f registry_version=NEW_UNPUBLISHED_REGISTRY_VERSION

The workflow requires that SHA to match its dispatch commit, rejects an existing registry version, repeats schema/npm/lifecycle checks, checks remote health/authentication denial, publishes through GitHub OIDC, and verifies the resulting record. It does not publish npm or rebuild the historical community release. If publication succeeds but final verification fails, inspect that exact registry version before retrying; existing records are never overwritten.

For an opt-in check through installed, already authenticated Codex and Claude Code CLIs, add --clients to the lifecycle command. This preflights account state and makes one bounded model request per authenticated client using only surveys_list and a temporary surveys:read credential for an empty local workspace. It does not purchase access, alter global client configuration, or run in ordinary CI. Each client has a three-minute timeout; unavailable authentication or approval is reported separately from the account-free protocol test. The optional run exits nonzero unless every selected client returns a verified empty tool result; use --clients=claude to retry only Claude after authentication. Client output is summarized without publishing raw logs or credentials.

On 13 September 2026, Codex CLI 0.154.0 completed a verified surveys_list call through public @likerts/[email protected] against a temporary scoped PostgreSQL workspace. Claude Code 2.1.201 parsed the configuration but remains unauthenticated, so its real-client call is still pending. See exact client evidence and the owner login action. The earlier Codex failure was traced to the test harness leaving stdin open; immediate EOF and reliable timeout tracking now have regression tests.

Source: tools/mcp/README.md at commit 3bfab27

Tools

0
SourceWeft has not read this server's tools yet.

Version history

1
  1. v0.1.1LatestSep 16, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listedRequired15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts