decisions-judge-mcp
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., "@decisions-judge-mcpGiven the attached state, judge whether we should merge this PR."
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.
decisions-judge-mcp
A small MCP server exposing a single judge tool: send natural language plus JSON application state, get back typed judgments — yes/no probability (noul), choice among options, or score on ordered levels — in one fast request, with model/usage metadata. Failures return a {fallback: true, error} envelope instead of blocking, so it is safe to compose into agent workflows.
Currently backed by the TypeSafe System One model (Jev) via @typesafe-ai/sdk. The provider-neutral judge primitive maps naturally onto related decision APIs such as OpenRouter alphadecisions.
Flagship consumer: clouatre-labs/agentic-coder-skill.
Tool: judge
Input:
state— JSON object or string (application state)questions— map of name →{type: "noul"|"choice"|"score", instructions, criteria?}noul: criteria omitted; answer is a yes/no probabilitychoice: criteria is{option: description|null}score: criteria is an ordered array of level descriptions
timeout_ms— optional, max 60000
Output: {answers, model, usage} on success, {fallback: true, error} on failure.
Related MCP server: jev-mcp
Install
Requires Node >= 20 and TYPESAFE_API_KEY in the environment (TYPESAFE_AI_TOKEN is accepted as a legacy fallback).
# recommended: run on demand via npx (no global install)
npx -y decisions-judge-mcp
# or pin a version for supply-chain reproducibility
npx -y decisions-judge-mcp@1.0.2
# or install globally
npm i -g decisions-judge-mcpTip: unpinned
npx -y decisions-judge-mcpalways runs the latest published version. For deterministic, supply-chain-hardened setups, pin an exact version or install globally and update deliberately.
MCP client configuration
All examples use npx -y, the standard pattern for Node-based MCP servers
(see Context7 and Brave Search). Swap in decisions-judge-mcp@<version> in the
args if you prefer pinning.
pi (~/.config/pi/mcp.json or equivalent):
{
"mcpServers": {
"decisions-judge": {
"command": "npx",
"args": ["-y", "decisions-judge-mcp"],
"env": { "TYPESAFE_API_KEY": "..." }
}
}
}Claude Code — one command:
claude mcp add --scope user decisions-judge \
--env TYPESAFE_API_KEY=... -- npx -y decisions-judge-mcpOr in ~/.claude.json:
{
"mcpServers": {
"decisions-judge": {
"command": "npx",
"args": ["-y", "decisions-judge-mcp"],
"env": { "TYPESAFE_API_KEY": "..." }
}
}
}goose (~/.config/goose/config.yaml):
mcp:
servers:
decisions-judge:
command: npx
args:
- -y
- decisions-judge-mcp
env:
TYPESAFE_API_KEY: "..."Windows note: some clients need cmd /c npx on Windows:
{
"command": "cmd",
"args": ["/c", "npx", "-y", "decisions-judge-mcp"]
}Development
npm ci
node --check server.mjs
node scripts/smoke.mjs # MCP stdio initialize handshake
node bin/decisions-judge-mcp # stdio server; needs TYPESAFE_API_KEY to answerLicense
Apache-2.0 — see LICENSE.
This server cannot be deployed
Maintenance
Related MCP Connectors
Tribeunal turns a question into a jury's verdict. An agent opens a case, a jury of humans and AI agents is seated, evidence is weighed and votes are cast, and the tally becomes a ruling the agent can long-poll for and act on. 39 tools, all annotated, plus eight Agent Skills that carry the procedure: how big a jury a decision needs, when a verdict actually lands, how to act on it. Arbitration mode bars the case owner from voting or closing early and enforces a quorum, closing with a verdict.
Pre-execution governance for AI agents. Deterministic PASS/FAIL/REVIEW verdicts, replayable proof.
Human judgment for AI agents: discover capabilities, get quotes, and track paid human tasks.
Preflight QA for AI-agent deliverables with structured verdicts and repair guidance.
Related MCP Servers
- AlicenseAqualityBmaintenanceEnables typed, calibrated judgment calls through classify, score, check, and batched ask tools, each returning full probability distributions for programmatic decisions.5220 npm4MIT
- AlicenseAqualityBmaintenanceEnables coding agents to make cheap, fast probabilistic decisions on every turn, with tools for coding-loop checks, review, verification, screening untrusted input, and ranking candidates.6220 npm7MIT
- AlicenseAqualityBmaintenanceEnables frontier coding agents to delegate routine probabilistic judgments to TypeSafe Jev, providing calibrated triage signals for failures, attempts, completion, context ranking, findings, risk, and generic evidence-grounded questions.7MIT
- AlicenseAqualityBmaintenanceEnables coding agents to compact conversation contexts verbatim, make fast decisions through choice, boolean, and rubric scoring, and enforce command safety guardrails.6MIT