Focus here to pause the changing names.

Paste this into your agent.

try-ennodia.cherninlab.com

Checkout still fails

  • Your chat · Codex

    retry.ts + test + earlier fixes

  • Ennodia → Claude Code

    reviews retry

    Compare with earlier fixes: Both reused the same session

Back in your chat: Retry uses an expired session

Test a fresh session on retry

View example run record

Illustrative bug investigation

Task
Explain why checkout still fails after two retry fixes.
Agent
Claude Code, called from the conversation in Codex.
Inputs
retry.ts, the failing test, and both attempted fixes.
Finding
The retry keeps using an expired session.
Comparison
Both earlier fixes changed the retry timing but kept the same session.
Next test
Retry with a fresh session and rerun the failing test.
Saved
The returned findings and task ID, alongside the attempts in your conversation.

This illustrates a workflow. The proposed fix still needs a passing test.

Examples

The bug survived two fixes. Now what?

Have another agent investigate the failing test and compare the attempted fixes in a separate session. Bring its findings back to your conversation.

Try this with your agent

Use your own failing test and attempted fixes. Paste this into the conversation where you’re working:

Use Ennodia MCP to investigate this bug with another available agent. Provide the failing test and our attempted fixes. Have that agent compare its explanation with those fixes and the source code in its own session. Have it identify the next test that can confirm or reject the explanation. Keep the investigation read-only. Bring a concise result and task IDs back here, including failed attempts and remaining uncertainty. We can inspect the captured details when needed.

Checkout still fails

  • Your chat · Codex

    retry.ts + test + earlier fixes

  • Ennodia → Claude Code

    reviews retry

    Compare with earlier fixes: Both reused the same session

Back in your chat: Retry uses an expired session

Test a fresh session on retry

View example run record

When your agent cannot listen

Give the excerpt to a model with native audio access through Ennodia. In this recorded check, Gemini returned the spoken words.

Try this with your agent

Use your own short audio excerpt. Paste this into the conversation where you’re working:

Use Ennodia MCP to check native listening on this short audio excerpt. Read the selected harness's input guidance first. Use one available harness and model, with comparison disabled. Have it report the file, range, tool, a few spoken words, and an audible observation, or the exact input error. Keep the task read-only. Do not substitute transcription or numeric metrics for native listening. Bring back the findings and task ID. Confirm access before planning a comparison of the full samples.

Can Gemini hear this file?

  • Your chat · Codex

    8-second MP3

  • Ennodia → Antigravity

    Gemini listens

    Reports listening evidence: File, range, spoken words

Back in your chat: Returns the spoken words

Next: compare matched samples

View recorded run

Catch the weak claim before the call

Review the meeting brief against the account notes. Compare the claims with their sources and turn missing information into questions for the call.

Try this with your agent

Use your own meeting brief and account notes. Paste this into the conversation where you’re working:

Use Ennodia MCP to review this meeting brief against the supplied account notes and product facts. Have another available agent compare the claims with their sources in its own session. Have it propose corrections and questions for the call. Keep it read-only and do not contact anyone. Bring a concise review and task IDs back here. We can inspect the captured details when needed.

Brief: budget approved

  • Your chat · Codex

    brief + notes

  • Ennodia → Claude Code

    checks brief

    Compare the two claims: Approval has no source

Back in your chat: Notes: budget proposal

Confirm budget on the call

View example run record

Compare skills without cluttering your conversation

Try the same task with and without a requested skill in separate conversations. Compare the findings and failed attempts before deciding what to use.

See both recorded attempts

Same code + same model

  • Plain review

    6 failing cases found

  • With review skill

    File read blocked

Findings kept

Skill benefit not established

Keep the tools you like

Ennodia connects the agents on your computer, using your models and subscriptions.

Check compatibility

Frequently asked questions

Is Ennodia another chat app?

Ennodia is an add-on for the agent you use. You stay in your usual conversation while it asks other installed agents to write code, answer questions, or review work.

What do I need to use it?

An app such as Codex that can connect to Ennodia, plus a supported agent installed on your computer. That agent needs a working account or API key. Your agent can use the setup address to check what’s missing and guide you through installation. Setup requirements.

Does it work with my subscriptions?

Ennodia uses each installed agent’s configured access. That can be a subscription or an API key, depending on the agent and provider. Their quotas and charges still apply. Ennodia itself is free.

What happens to my code and files?

Ennodia runs locally. It passes the requested task to your selected agents, which can send material to their configured providers. There is no Ennodia-hosted service. Local run history can be disabled.

What if every agent gets stuck?

You have a record of the approaches tried and the evidence they produced. Use that to reconsider the task or test another assumption. Agreement between agents does not prove correctness or impossibility.

Will it always be faster or cheaper?

Additional agents consume resources. The aim is to reduce manual switching and repeated attempts across the whole task. We publish examples and recordings, but have not established general time or token savings.

Try Ennodia on the task you have open right now

Paste this into your agent.

try-ennodia.cherninlab.com

Use your preferred compatible chat app, CLI, or IDE.
Your agent gets the instructions to install and connect Ennodia.

Manual setup

If Ennodia fits your workflow, give it a star on GitHub.