Skip to content

INTEGRATIONS · HYVE FROM INSIDE YOUR IDE

Use HYVE without leaving your editor.

HYVE MCP Bridge exposes Omega, Narrator, Augur, Trust, and Leaper as native MCP tools. Drop the manifest into Claude Code, Cursor, Zed, or Continue.dev and your AI assistant calls HYVE organs directly — no context switch, no terminal, no copy-paste.

HYVE MCP Bridge

A standalone binary that exposes Omega, Narrator, Augur, Trust, and Leaper as native MCP tools — call HYVE organs directly from Claude Code, Cursor, Zed, and Continue.dev.

Five organs. One bridge.

Each organ is a single MCP tool. The bridge handles authentication, rate-limiting, and result-shaping; your IDE agent just sees five simple capabilities it can compose.

Omega

decide() function — refuse / clarify / proceed gate over 4 signals.

hyve-mcp-bridge omega decide ...

Narrator

Render any structured event into 5 audience tones (SOC, Executive, Legal, Helpdesk, MSSP).

hyve-mcp-bridge narrator render --audience executive ...

Augur

Static analysis security gate. 11 detectors, defensive wire format.

hyve-mcp-bridge augur audit . --fail-on-findings

Trust

Reputation lookup with EWMA decay + cross-tenant federation.

hyve-mcp-bridge trust query --agent-id X --tenant Y

Leaper

Run Windows apps (via Wine), containers, Flatpak apps and retro game ROMs.

hyve-mcp-bridge leaper run path/to/app.exe

Per-harness skill manifests, included.

Per-IDE skill manifests (.claude/skills, .cursor/rules, ~/.continue/config.json) so operators invoke HYVE organs from inside their AI coding harness without leaving the editor. Click any harness below for the ready-to-paste configuration.

Claude Code

.claude/skills/
SHOW MANIFEST ▾
# .claude/skills/hyve-omega.md
---
name: hyve-omega
description: HYVE Omega 4-signal decision orchestration — call decide() with tide/trust/narrator/augur signals
---

When the user asks the agent to make a decision under uncertainty, call:

```bash
hyve-mcp-bridge omega decide --tide "$T" --trust "$R" --narrator "$N" --augur "$A"
```

Returns one of: refuse / clarify / proceed.

Cursor

.cursor/rules/
SHOW MANIFEST ▾
// .cursor/rules/hyve-augur.mdc
---
name: hyve-augur
trigger: on-save
---

Before committing any code to repository, run:

`hyve-mcp-bridge augur audit --severity high --fail-on-findings`

Block the commit on any HIGH or CRITICAL finding from the 11 detectors
(secrets, weak_crypto, sql_injection, command_injection, deserialization,
insecure_random, unsafe_audit, panic_audit, xss, regex_dos, ssrf).

Zed

settings.json
SHOW MANIFEST ▾
// Zed settings.json
{
  "tools": {
    "hyve": {
      "command": "hyve-mcp-bridge",
      "args": ["serve", "--port", "7900"],
      "transport": "stdio"
    }
  }
}

Continue.dev

~/.continue/config.json
SHOW MANIFEST ▾
// ~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "hyve",
      "command": "hyve-mcp-bridge",
      "args": ["serve", "--mode", "mcp"]
    }
  ]
}

Skills + Closed Beta

The bridge runs locally on localhost:7900 — the bridge itself sends nothing off your machine, though the IDE assistant you call it from may use its own cloud service.

Request Beta Access