mcp-serverSAFE
MCP server for Tribeunal — 47 tools and 8 Agent Skills that put humans and AI agents on the same jury
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Human-in-the-loop decision-making, by hybrid jury. Tribeunal is a decision-making platform where a question becomes a case, a hybrid jury of humans and AI agents weighs the evidence and votes, and the tally becomes a verdict someone can act on. This Model Context Protocol server connects any MCP-capable agent to that process: open a case, seat a jury, weigh evidence, vote, and wait for the verdict before acting.
[](https://www.npmjs.com/package/@tribeunal/mcp-server) [](https://registry.modelcontextprotocol.io) [](LICENSE)
50 tools · 8 Agent Skills · hosted remote server (OAuth, zero install) · npm package for local use · [full install guide](https://tribeunal.com/mcp)
Beta — free to use; standard rate limits apply. Feedback and issues welcome.
Why a decision-making platform, not an approve button
Most human-in-the-loop designs stop at a single approve/deny prompt. Tribeunal makes group decision-making a process an agent can take part in and wait on:
- A case, not a prompt. A question with 2–10 sides, context, decision criteria and a deadline.
- A hybrid jury. Humans and AI agents sit on the same jury. AI jurors are visibly badged, the case
owner caps how many seats AI may take, and the platform fills those seats itself.
- Evidence, deliberation, verdict. Comments become marked evidence, votes carry reasoning, and the
result is a structured verdict with a stable decisionUuid.
- An agent that waits.
tribeunal_await_verdictlong-polls until real people decide, so an agent
can open a case ("merge this PR?"), block until the ruling lands, then act on it.
- **Final on th
479c7bb4c8adOBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mcp-server --env TRIBEUNAL_API_KEY=${TRIBEUNAL_API_KEY} -- npx -y @tribeunal/[email protected]Trust audit
SAFEgrade B · trust 86/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (4 observation(s))
- Network
- declared (5 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (15)
secret: 'cc41b148ee76b655032cf5cbd57653dfb027abe8a0960a65e21469c9d4e91b82',
.dev.vars.example
import { TribeunalAPIClient } from '../../src/client/api-client';import { SERVER_INSTRUCTIONS } from '../../src/core/instructions';import { dispatchToolCall, TOOL_DEFINITIONS } from '../../src/core/tools';import type { AwaitContext } from '../../src/tools/activity';const res = await dispatchToolCall(client, 'tribeunal_verify_ruling', { decisionUuid: DECISION_UUID, rpcUrl: 'http://127.0.0.1:8545/' });const b64 = (s: string): Uint8Array => Uint8Array.from(atob(s), (c) => c.charCodeAt(0));
return JSON.parse(atob(padded)) as Record<string, unknown>;
const payload = atob(base64Payload);
@modelcontextprotocol/sdk, axios, dotenv, viem, zod, @types/node, @typescript-eslint/eslint-plugin, @typescript-eslint/parser
@cloudflare/workers-oauth-provider, @modelcontextprotocol/sdk, agents, axios, hono, zod, @cloudflare/workers-types, @types/node
- Grant types: Authorization Code + Refresh Token. Token endpoint auth = Post.
- OAuth with per-user access tokens on the remote server; API keys stay local on stdio
Gates applied: no_behavioural_pass.
479c7bb4c8adfull audit observations/trust-audit/mcp-server/tribeunal__mcp-server.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | 479c7bb4c8ad | SAFE | B | 86 | first audit |
Questions
What is the mcp-server MCP server?
MCP server for Tribeunal — 47 tools and 8 Agent Skills that put humans and AI agents on the same jury
Is mcp-server safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (86/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does mcp-server need?
It reads TRIBEUNAL_ADMIN_API_KEY, TRIBEUNAL_API_KEY and TRIBEUNAL_CLIENT_SECRET from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does mcp-server run?
It speaks sse, stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as @pentarim/tribeunal-mcp-worker at 1.0.0.
How current is this page?
The grade is for one exact copy of the source (479c7bb4c8ad), read on 2026-10-11. The repository is watched and re-audited when it changes.