Hrevn Mcp Server

io.github.ai-human-andalusiaPublished by GitHub account ai-human-andalusiav0.1.0a0Updated Oct 24, 2026

HREVN trust layer for baseline diagnostics and verifiable workflow integrity for AI agents.

VerifiedSTDIODesktop onlyDeveloper ToolsSecurity & Monitoring

Overview

AI-generated overview

A thin local MCP server that exposes HREVN baseline diagnostics and workflow-integrity checks as tools backed by the hosted HREVN API.

What you can do
Run a baseline check on a task record and profile, returning structured results with risk flags, remedy payloads and check IDs. Validate a record against a named profile. Generate a bundle from a record and traces, optionally including a report PDF. Verify an existing bundle or artifact from a local path. All canonical logic stays in the remote HREVN runtime; the server only exposes stable tool contracts.
When to use it
Use it when you want an assistant to run HREVN baseline diagnostics, profile validation or bundle generation and verification as part of an AI workflow, and you already have access to the HREVN managed API.
Requirements
Runs as a local Python process, installed from PyPI or from source; the console entry point is hrevn-mcp-server over stdio. Needs network access to the HREVN API and an issued API key supplied via the HREVN_API_KEY environment variable, plus HREVN_API_BASE_URL. Optional timeout and server-name/version variables exist.
Before you install
Requires a HREVN API key, which is passed in the client configuration environment; keep it out of shared configs. The self-test and baseline check call the live managed API, so task records and traces are sent to a third party. Bundle generation can produce report PDFs, and verify_bundle reads a local path you specify.

Installation

In SourceWeft

  1. Open Hrevn 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.

Other MCP clients

Follow the launch instructions in the repository.

README

HREVN MCP Server

Minimal stdio MCP server for HREVN, backed by the live managed runtime at https://api.hrevn.com.

Why this exists

This repo exposes HREVN as real MCP tools instead of ad hoc helper scripts. It stays intentionally thin:

  • canonical semantics remain in the HREVN managed runtime
  • the MCP layer only exposes stable tool contracts
  • consequential results come back as structured output, including risk_flags, remedy_payload, and check_id

Included tools

  • baseline_check
  • profile_validate
  • generate_bundle
  • verify_bundle

Environment

bash
export HREVN_API_BASE_URL="https://api.hrevn.com"export HREVN_API_KEY="replace-with-issued-alpha-key"

Optional environment variables:

  • HREVN_MCP_TIMEOUT_SECONDS
  • HREVN_MCP_SERVER_NAME
  • HREVN_MCP_SERVER_VERSION

Install

Editable install

bash
cd hrevn-mcp-serverpython3 -m pip install -e .

From PyPI

bash
python3 -m pip install hrevn-mcp-server

The console entry point is:

bash
hrevn-mcp-server

Run

bash
hrevn-mcp-server

The server uses MCP stdio transport.

Verify before using a client

bash
hrevn-mcp-server --versionhrevn-mcp-server --list-toolshrevn-mcp-server --self-test

If you are running directly from source without an installed entry point:

bash
PYTHONPATH=src python3 -m hrevn_mcp_server.server --list-toolsPYTHONPATH=src python3 -m hrevn_mcp_server.server --self-test

The self-test runs a live baseline_check against the configured HREVN managed API.

Tool contracts

baseline_check

Minimal payload:

json
{  "task_type": "ai_workflow",  "profile": "eu_readiness_profile",  "record": {    "agent_name": "example_agent",    "summary": "baseline smoke test"  },  "metadata": {    "surface": "mcp",    "stage": "pre_completion"  }}

profile_validate

Minimal payload:

json
{  "profile": "eu_readiness_profile",  "record": {},  "metadata": {}}

generate_bundle

Minimal payload:

json
{  "record": {},  "traces": [],  "options": {    "include_report_pdf": false  }}

verify_bundle

Minimal payload:

json
{  "source": "/path/to/bundle-or-artifact"}

Example MCP config

Use a local stdio configuration like this:

json
{  "mcpServers": {    "hrevn": {      "command": "python3",      "args": [        "-m",        "hrevn_mcp_server.server"      ],      "env": {        "PYTHONPATH": "/ABSOLUTE/PATH/TO/hrevn-mcp-server/src",        "HREVN_API_BASE_URL": "https://api.hrevn.com",        "HREVN_API_KEY": "YOUR_HREVN_API_KEY"      }    }  }}

See also:

Antigravity status

HREVN can already be explored experimentally in Google Antigravity through custom MCP configuration.

What is already validated:

  • the HREVN MCP server runs locally
  • the server exposes real tools backed by https://api.hrevn.com
  • baseline_check returns real structured results
  • Antigravity preview builds expose MCP configuration surfaces in the UI

What is not claimed yet:

  • official native Antigravity integration
  • guaranteed MCP tool injection into every Antigravity agent instance
  • Antigravity-specific guardrail hooks beyond what has been directly observed

Registry metadata

This repo includes:

as a machine-readable manifest describing the MCP server, package, transport, and repository metadata.

Design rule

This server must not reimplement HREVN truth locally. It should expose stable MCP tools that call the managed API.

Source: README.md at commit 234f04b

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.0a0LatestSep 16, 2026

Score

40% 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

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • 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 listed15 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

    Declare a license in the repository or the registry entry.