Documentation

API Keys

How does Metamorfon work with API keys?

Metamorfon works on a BYOK (Bring Your Own Key) model: you use your own API keys from the model providers you want to connect.

In practice, you create your keys directly with the relevant providers, and all API usage billing (tokens, quotas, subscriptions, rate limits) is handled exclusively by those providers and under your own accounts.
Metamorfon does not charge for API calls. The application simply displays usage indicators in the interface, such as token consumption and estimated costs, to help you monitor your usage.

Where to enter your keys

Go to Settings > API Keys.

For each provider, you can:

• paste your API key;
• show or hide it;
• validate it to check whether it is accepted;
• retrieve the list of available models for your account after validation.

Storage and privacy

Your API keys are stored locally on your device — in the browser or in the desktop application, depending on how you use Metamorfon.
They are not sent to a remote Metamorfon server.

They are used only to allow the application to communicate with the official APIs of the providers you have configured.

For key validation and model list retrieval, the application relies on a local backend.

Local backend and validation

Metamorfon uses a local server (by default: http://127.0.0.1:8000) to validate API keys and retrieve available models from providers.

The application attempts to start this backend automatically.
If validation fails or the server appears unavailable:

• wait a few seconds;
• click Retry in the API Keys section;
• check that no other program is using port 8000;
• consult the documentation or the logs referenced in the application if the issue persists.

Supported providers

Metamorfon currently supports several model providers, including:

• OpenAI
• Anthropic
• Google
• Mistral
• xAI
• DeepSeek
• Cohere

Each API key must be generated and managed through the official interface of the relevant provider.

Common error messages

Invalid or expired key, or no model available

Check that the key is correct, linked to the right account or project, and that at least one compatible model is enabled for your subscription or region.

Quota exceeded or rate limit reached

Check the provider’s dashboard for billing status, usage limits, or service availability.

Local backend unavailable

Make sure the local server is running correctly, that port 8000 is free, and then try validation again.

Security best practices

• Never share an API key in plain text.
• Do not expose your keys in screenshots, forums, or support tickets.
• If a key is compromised, revoke it immediately from the provider’s website and generate a new one.
• When possible, use keys with restricted permissions.

For advanced users

In custom setups — especially if you run the backend yourself — environment variables such as OPENAI_API_KEY, ANTHROPIC_API_KEY, or GOOGLE_API_KEY may complement or replace manual entry in the interface.

For standard use through the graphical application, the Settings > API Keys section is sufficient.

Quick Start

1. Install and launch Metamorfon on a supported system (Windows or macOS).
2. Wait until the local backend is running (see Local backend below). The app usually starts it for you; status is shown in the interface.
3. Open Settings → API keys, add at least one key for a provider you use, then validate the key and sync models so the model list is up to date.
4. Create a new session, choose your dialogue strategy and models, enter your prompt, and run the first turn.
5. Use export when you want a portable copy of the session (Markdown, Word, or JSON snapshot—see Sessions, export & import).
   If something fails at step 2 or 3, see Troubleshooting.

Account & Subscriptions

Local Backend

Metamorfon relies on a local FastAPI server (by default at http://127.0.0.1:8000) for tasks such as API key validation, listing available models, and orchestrating calls during use. The application tries to start this server automatically.

If validation or model sync fails:

• Wait a few seconds and use Retry in the API keys area.
• Ensure nothing else is using port 8000 on your machine.
• Check that your firewall or security software is not blocking local traffic to that address.
• If problems persist, use any log paths or diagnostics shown in the app, or follow your internal runbook for starting the backend manually (advanced setups).
• The backend is local; it is not the same as sending data to Metamorfon’s website.

Privacy & data flow

API keys are stored on your device. They are not sent to Metamorfon’s servers on the internet. They are used so the application can call your chosen providers’ official APIs according to their terms.

Prompts, completions, and related content are processed according to how you configure the session (models, web access, etc.). In normal use with cloud models, that content is sent to the LLM providers you selected—not to Metamorfon as a central processor of your conversations. For precise legal wording (retention, subprocessors, regions), refer to your Privacy Policy and keep this section aligned with it.

Exports may contain full session text; treat shared files like any sensitive document (see Sessions, export & import).

Sessions, export & import

Export lets you take the dialogue out of the app in several formats, for example Markdown (portable, easy to read in any editor), Word, or a JSON snapshot of session state.

Import (where available) restores a session from a JSON snapshot so you can resume with the same parameters and history, depending on version compatibility.

Before sharing an export: remove or redact sensitive content; exported files are not automatically anonymized.

Metrics & costs

The interface shows usage indicators such as tokens and estimated costs so you can monitor consumption during a session. These figures are meant to help you track usage in the app; they may not match your provider’s invoice line-by-line (rounding, pricing tiers, cached tokens, or provider-specific billing rules).

Billing for API calls is between you and each LLM provider under your account (BYOK). Metamorfon does not charge you for provider API usage.

Important Settings (overview)

• Temperature (per model): Adjusts randomness versus determinism; lower values tend to be more focused, higher values more varied.
• Web access (when available): Allows the session to use live web sources where supported. This can affect latency, cost, and what data leaves your machine (requests may go to the provider’s web-enabled pipeline). Enable only when you need it.
• Default models (e.g. for synthesis or summarization): Set sensible defaults in Settings so you do not have to reconfigure every session.
• Dialogue strategies & debate modes: These shape how models take turns and how strongly they challenge each other. Short explanations in-product help.

Troubleshooting

System requirements & environment

• OS: Windows and macOS builds as offered on the site.
• Network: Required for cloud LLM APIs and optional web features; local backend uses localhost.
• Corporate networks: If API calls fail only on office Wi‑Fi, ask IT about TLS inspection, proxy, or egress rules for provider domains.