MCP.Market

Proof Layer MCP server

by World-999-Labs·io.github.World-999-Labs/proof-layer·v1.0.1
No reviews yet

Signed, hash-chained provenance receipts for AI agent actions.

B71/100grade B
Write a reviewInstall
What users say
No reviews yet
Be the first
Safety scan
B71/100

full report

Adoption
New

0 stars

Reviews

Write one

Nobody has reviewed Proof Layer yet.

If you have run it, two minutes of your experience saves the next person an afternoon.

Proof Layer tools (8, 2 write)

write = sends, deletes, buys or posts

Read from the package source without running it. The installed server may list more.

  • pl_create_planwrite action

    A human approves a governed plan \u2014 a scoped, time-bounded set of future actions across named agents. Returns a plan_id and JTI that agents use to request delegation receipts via pl_delegate. Scope patterns support wildcards: 'build:*' authorizes any action starting with 'build:'.

  • pl_delegate

    An agent requests authorization to perform a specific action under a parent plan. Returns 'ok' with a signed receipt (proceed with the action), 'checkpoint_required' (human must re-approve this specific action), or 'denied' (agent or action not authorized under the plan).

  • pl_evaluatewrite action

    Submit an AI agent action proposal to Proof Layer for governance evaluation. Returns a cryptographically-signed receipt with a verdict of EXECUTE, BLOCK, REVIEW, or SHADOW. Use this before any consequential agent action (sending emails, modifying data, making payments, etc.).

  • pl_get_receipt

    Retrieve a Proof Layer audit receipt by its UUID. Returns the full receipt including verdict, signature, and policy pack hash.

  • pl_list_policy_packs

    List all governance policy packs for the authenticated tenant, including which one is currently active.

  • pl_list_receipts

    List recent Proof Layer audit receipts for the authenticated tenant. Supports filtering by agent and verdict.

  • pl_usage

    Get current Proof Layer quota usage \u2014 receipts used, receipts remaining, credit balance, billing cycle dates, and plan tier.

  • pl_verify_receipt

    Cryptographically verify a Proof Layer receipt. Checks the Ed25519 signature and confirms the signing key is ACTIVE. Returns valid: true/false with a reason.

Public scan report

scanner v0.1.9 · 2026-09-27 · same rubric, same numbers if you re-run it

no findings
  • Code scan2 source files scanned25/25
  • –Live reliabilityno gateway calls yet and no remote to proben/a
  • –Tool poisoningtools not inspected (local package is not executed); not countedn/a
  • Auth qualitystatic API keys via environment variables6/15
  • Maintenancelast push 147 days ago8/15
  • Maintainer identityregistry namespace matches repository owner7/10
Overall 71/100. Components that don't apply are left out of the denominator. Any critical finding is an F.RubricAppeal a findingJSON

How Proof Layer compares

Against every graded AI and agents MCP server, from our own scans and uptime checks. Updated after each scan run.

Safety score
71/100
higher than 38% of 4,460 AI and agents servers
Tools
8
the typical AI and agents server has 6
Tools that change things
25%
typical: 0%. Fewer write tools means less an agent can break

What its tools let an agent do: 6 read · 2 change

Install Proof Layer in Claude Code, Cursor or VS Code

Runs npx -y @proof-layer/mcp on your machine. Read the scan report first; the gateway never runs local packages.

claude mcp add proof-layer -- npx -y @proof-layer/mcp
Add to Cursor

What the publisher says

From the Proof Layer repository's README, as published. We do not edit it. Read it on GitHub

Proof Layer

Cryptographic governance receipts for AI agents. Issued before the agent acts. Ed25519-signed. Hash-chained. Verifiable offline.

Built by WORLD999LABS. Published on npm under @proof-layer/. The legacy @veridocs/* package names continue to resolve and forward to the new ones.

What this repo contains

This is the public half of Proof Layer — the parts you install and run on your own machine:

The kernel (governance engine, Gauntlet adversary/judge pipeline, billing, dashboard) is closed-source and runs at prooflayer.world999labs.com. Open-source clients + closed-source server is the same pattern Stripe, Resend, and Vercel use — you get full transparency on what runs locally and what data leaves your machine, while the proprietary server logic stays protected.

Quick start

npx -y @proof-layer/mcp@latest

Drop into Claude Desktop's claudedesktopconfig.json:

{
  "mcpServers": {
    "proof-layer": {
      "command": "npx",
      "args": ["-y", "@proof-layer/mcp@latest"],
      "env": {
        "PROOF_LAYER_API_KEY": "pl_live_...",
        "PROOF_LAYER_API_URL": "https://prooflayer.world999labs.com"
      }
    }
  }
}

Restart Claude. Eight pl_* tools appear in the picker. Get an API key at prooflayer.world999labs.com.

Migrating from @veridocs/mcp? No code changes required. The legacy package name continues to resolve, the legacy vd tool names continue to work alongside pl, and existing VERIDOCSAPIKEY / VERIDOCSAPIURL environment variables are still accepted. You can switch the package name and env-var names on your own schedule.

Why pre-execution governance?

Most "AI safety" tools run after the model has acted — toxicity filters, output classifiers, post-hoc audit logs. By the time the alarm fires, the agent has already sent the email, merged the PR, or executed the trade.

Proof Layer flips this:

  1. Agent proposes an action (e.g. "send wire transfer for $50,000")
  2. Proof Layer evaluates → returns signed verdict: EXECUTE / BLOCK / REVIEW / SHADOW
  3. Receipt is written to a hash-chained, Ed25519-signed audit trail
  4. Anyone with our public key can verify the entire chain offline — no trust in our infra required

A broken hash chain = tampering, immediately visible. A missing receipt = the agent acted without permission.

Verifying receipts offline

Receipts are designed to outlive us. Here's how to verify a bundle without ever calling our API:

import { verifyBundle } from "@proof-layer/verify";
import fs from "fs";

Shortened. The full README is on GitHub.

Nothing above is checked by us. What we check is on the safety report.

Proof Layer: common questions

Is Proof Layer MCP server safe?
Mostly: it is graded B (71/100). Read the Proof Layer safety report
How do I install Proof Layer?
It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
Does Proof Layer need an API key?
Yes. The registry entry asks for PROOF_LAYER_API_KEY.
Is Proof Layer maintained?
The last commit was 184 days ago (2026-05-03). The latest release is v1.0.1.
What can I use instead of Proof Layer?
Servers from other publishers that do the same job: Emilia Protocol MCP server, three.ws Provenance MCP server and agent-device MCP server.
How does Proof Layer compare with other AI and agents MCP servers?
safety score 71/100 (higher than 38% of 4,460 AI and agents servers); tools 8 (the typical AI and agents server has 6); tools that change things 25% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.

Alternatives to Proof Layer

Same job from other publishers: the closest match first, then the best rated.

  • Emilia Protocol
    Exact-action approval for consequential agent actions: request, track, and verify signed receipts.
    B
  • three.ws Provenance
    Append-only, signed, on-chain-verifiable agent action log — record and audit what agents did.
    More from nirholas →
    B
  • agent-device
    Open source device layer for AI agents: control mobile, web, TV, and desktop apps over MCP
    A
  • Decionis
    Authorize consequential AI agent actions before execution
    A
  • Agent-Native Chat
    Minimal chat-first app with durable threads, actions, and the app-agent loop
    A