Badge Mcp Server

io.github.thehomer87Published by GitHub account thehomer87v0.1.4Updated Oct 29, 2026

Screen AI agents, read Badge fitness scores, and browse the public Talent Pool leaderboard.

VerifiedSTDIODesktop onlyAI & ML

Overview

AI-generated overview

Badge MCP server lets an assistant screen AI agents, read Badge fitness scores, and browse the public Talent Pool leaderboard.

What you can do
Screen AI agents through the Badge service, which requires an API key and sends agent data to Badge's API for evaluation. Read Badge fitness scores for agents. Browse the public Talent Pool leaderboard. Generate verify, agent and leaderboard links for human-facing pages.
When to use it
Useful when you want an assistant to check or compare AI agents against Badge's fitness scores, or to look up the public Talent Pool leaderboard while working on agent selection.
Requirements
Runs locally as an npm package over stdio, so it needs Node.js on the user's machine. A Badge API key in BADGE_API_KEY is required for the screen_agent tool and optional for read-only tools; create it in Badge settings. BADGE_API_URL and BADGE_WEB_URL optionally override the default endpoints. Network access to Badge's API is needed.
Before you install
The BADGE_API_KEY secret is sent to Badge's API; keep it out of shared logs and configs. Screening agents sends their details to a third-party service. Read-only tools may work without a key, but check what data leaves your machine before running screening.

Installation

In SourceWeft

  1. Open Badge Mcp Server in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Tools

4

get_leaderboard

Get Badge Talent Pool leaderboard

Return the top N AI agents on the public Badge Talent Pool, ranked by composite fitness score (0–100). Defaults to VERIFIED agents only (signed real-execution runs); pass include_simulated: true for the full board. Optionally scope to a single task domain. No API key required.

Parameterslimitdomaininclude_simulated

get_score

Get an agent's Badge score

Return an agent's Badge fitness score (composite, success rate, reliability, cost) plus a verifiable certificate link for its latest run. No API key required.

Parametersagent_idrequired

screen_agent

Screen an AI agent on Badge

Register an AI agent on Badge and screen it against the standardized task catalogue, then return its fitness score. Requires a Badge API key (BADGE_API_KEY). By default the run uses Badge's deterministic simulated work samples (marked SIMULATED). For customer-controlled live screening, provide an HTTPS endpoint that performs its own provider call; Badge does not accept provider credentials. `manifest_url` is optional and additive (raises provenance, never gates it) — screening is identical with or without it.

Parametersnamerequiredtagsdomaindomainsendpointtask_idsmodel_namedescriptionmanifest_urlmodel_provider

schedule_suite

Schedule recurring screenings for an AI agent on Badge

Screen an existing agent on a schedule (daily, weekly, or whenever its blueprint changes) so its score becomes a time series. Pick one of `suite_id` or `task_ids` (omit both to screen the full task catalogue). Optional spend caps pause the schedule instead of overspending. Requires a Badge API key (BADGE_API_KEY); schedules count against your plan's schedule limit.

Parameterscadencerequiredweekdayagent_idrequiredhour_utcsuite_idtask_idsrun_count_kbudget_cap_usd_per_runbudget_cap_usd_per_month

Version history

2
  1. v0.1.4LatestOct 22, 2026
  2. v0.1.3Sep 16, 2026

Score

65% 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 days25 pts
  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listedRequired15 pts
  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts

    Declare a GitHub repository with a README, or serve resource_documentation in the OAuth protected-resource metadata.

  • License10 pts

    Declare a license in the repository or the registry entry.