Ennodia

What is Ennodia?

Ennodia is an MCP (Model Context Protocol) server that lets a primary agent orchestrate reviews across multiple installed local agent CLIs, track every child task, and use a model-led Compare pass to surface agreements, disagreements, blind spots, and a single synthesized answer with traceable receipts. It is designed for deliberation-class work where no single model should be the only reviewer.

How to use Ennodia?

Send the link try-ennodia.cherninlab.com to your primary agent and let it handle setup, or run the server directly with npx -y ennodia (requires Bun 1.3.14+). The main MCP tool is ennodia_run; use ennodia_get_run to poll results. For local HTTP/TypeScript access, use the separate package @cherninlab/ennodia-io via npx -y @cherninlab/ennodia-io.

Key features of Ennodia

• Discovers available local AI tools and plans a review route.
• Estimates preflight input tokens and enforces local caps.
• Starts, monitors, and cancels child agent tasks.
• Compares multiple completed outputs and synthesizes one answer.
• Returns traceable receipts with timing, logs, and failures.
• Supports eight agent harnesses including Codex CLI and Claude Code.

Use cases of Ennodia

• Multi-model code review: ask several agents to review the same diff and compare results.
• Red-teaming: use multiple agents to find blind spots in a design or specification.
• Decomposing complex tasks: split work among specialized agents and synthesise output.
• Audit trails: generate traceable receipts for every child task and comparison pass.

FAQ from Ennodia

What runtime does Ennodia require?

Ennodia requires Bun version 1.3.14 or newer and is typically launched via npx -y ennodia or bunx ennodia.

What agent harnesses does Ennodia support?

Supported harnesses include Codex CLI, Claude Code, OpenCode, Kilo Code, Kiro CLI, Cline CLI, Hermes Agent, and Antigravity.

How does Ennodia compare outputs?

After child agents finish, Ennodia runs a model-led Compare pass that surfaces agreements, disagreements, blind spots, and synthesizes one answer with traceable receipts.

Is there a way to use Ennodia outside of MCP?

The separate @cherninlab/ennodia-io package exposes a local HTTP and TypeScript interface for apps that want BYOK-style settings over installed local agents.

Where does Ennodia store data?

Child task state, logs, and comparison data are stored locally; the project’s data governance documentation covers local storage and data movement boundaries.