Proxy Account Format Converter

What Is the Proxy Account Format Converter?

The Proxy Account Format Converter is a free browser tool that translates AI relay and gateway account files from one format into another. Different proxy and management tools each expect their own JSON shape: CLIProxyAPI reads one auth file per account, Sub2API wants an import envelope, 9Router has its own bulk structure, and Codex, Claude Code, Gemini CLI and Qwen Code each keep a native local credential file. Moving accounts between them by hand means editing keys and nesting structures until something breaks. This converter does it for you: you paste or drop your account JSON, it detects the source format and the provider of every account, and it rebuilds them in the format you picked. All of the parsing happens in your own browser with JavaScript, so the tokens in those files never travel anywhere.

Why Use It and Who It Is For

Anyone who runs more than one AI gateway ends up with account files in several formats. The converter exists for that exact situation.

• Relay operators who are switching from one proxy tool to another and do not want to re enter every account by hand.
• Developers testing several gateways who keep the same set of OpenAI, Claude, Gemini and Antigravity accounts across tools.
• People importing a shared account dump who need it reshaped before the target tool will accept it.
• CLI users moving Codex, Claude Code, Gemini CLI or Qwen Code credentials between machines and formats.
• Anyone concerned about privacy who refuses to paste access tokens into an unknown web service.
• Teams cleaning up duplicates before a bulk import, so the same account does not appear twice.

How This Tool Works

Step one reads whatever you give it. The input box accepts a single JSON document, several documents, JSON Lines, arrays and nested exports, and the file area accepts multiple .json or .txt files at once. The parser walks every object it finds and matches it against known shapes, so a mixed paste of a ChatGPT session export and a Codex auth file is sorted out automatically. When an object carries no provider marker - a bare OAuth token, for example - you can set the provider manually in Provider for undetected accounts, otherwise it is skipped and reported.

Where the file stores an email, account id, plan or expiry only inside a JWT, the tool decodes the access and ID tokens to recover those values instead of leaving them blank. It also reads expiry fields such as expires_at, expiry_date and expired so the results table can show you which accounts are stale.

Step two rebuilds the accounts for your chosen target. Each target supports a specific provider list, and anything outside that list is skipped rather than silently corrupted. Two options shape the output: Remove duplicate accounts merges repeated accounts and keeps the more complete record, and Differentiate account priority gives accounts imported first a higher usage priority in the generated file. Step three shows parsed account counts, ready counts, warnings and skipped items, then lets you copy the JSON, download it, or - for targets that need one file per account - download a .zip laid out for the target tool directory. Download all formats produces one archive with a folder per supported target.

Key Features

• Eleven import sources: ChatGPT session, Codex auth.json, CPA (CLIProxyAPI), Sub2API, Cockpit Tools, 9Router, AxonHub, Codex-Manager, Claude Code, Gemini CLI and Qwen Code.
• Ten target formats: Sub2API, CPA, Cockpit Tools, 9Router, Codex auth.json, AxonHub, Codex-Manager, Claude Code, Gemini CLI and Qwen Code.
• Ten recognized providers: OpenAI/Codex, Claude, Gemini, Antigravity, Qwen, iFlow, Kimi, xAI/Grok, Kiro and GitHub Copilot.
• Automatic format and provider detection across mixed and multi-document input.
• Multiple file upload with drag and drop for .json and .txt.
• Manual provider override for undetected accounts.
• Duplicate merging and optional priority ordering.
• Warnings for expired access tokens, missing refresh tokens and accounts the target cannot accept.
• Per-account files packed into a .zip automatically, plus a download of every format at once.
• Built in example data so you can try the whole flow before importing anything real.
• Runs entirely in the browser: no upload, no server logging, no browser storage of your accounts.

How to Use This Tool

1. In Import account JSON, paste one or more JSON documents or JSON Lines into the text area, or click the file area to add .json and .txt files. Load example fills the box with sample accounts and Clear all empties it.
2. If some accounts have no provider marker, choose one in Provider for undetected accounts. Leave it on auto detect if your input already names the provider.
3. In Choose target format, pick the tool you want to import into. The provider list on each target tells you what it accepts.
4. Turn on Remove duplicate accounts to merge repeats, and Differentiate account priority if the target should favour the accounts you imported first.
5. In Review and download, check the counters for parsed accounts, ready accounts, warnings and skipped items, and read the results table with provider, account, detected source, expiry and target columns.
6. Use the output file selector if the target produced more than one file, then click Copy JSON, Download current format or Download all formats.

Limits and Rules

The target format decides what can be converted. Codex auth.json and Codex-Manager accept OpenAI accounts only, Claude Code accepts Claude only, Gemini CLI accepts Gemini only, and Qwen Code accepts Qwen only. An account is skipped when the target does not support its provider, when no access token, refresh token or API key is present, or when the provider cannot be detected and you did not choose one. Warnings are not failures: an expired access token is still converted, and the target tool can refresh it if a refresh token came along.

There is no server side size limit because there is no upload, but very large pastes are limited by your browser memory, so split huge collections into batches. Access and refresh tokens are live credentials - anything you download can sign in as those accounts. Store exported files like passwords, never commit them to a repository, never publish them, and delete the archive when you are done. Use the tool only with accounts you own or are authorised to manage.

When to Use It and When Not To

Use it when you are migrating between relay tools, when an import refuses a file it cannot recognise, when you want one tidy archive of every format for backup, when a shared dump needs duplicates removed before import, or when you are setting up a CLI credential file on a new machine.

Do not use it to obtain accounts you are not authorised to use, to store tokens long term, or as a password manager. It cannot refresh a token that has no refresh token, it cannot convert a provider into a target that does not support that provider, and it cannot fix a file that is not account JSON in the first place. If the parsed count stays at zero, the input is not in a recognised format - use Load example to see what a valid file looks like.

Frequently Asked Questions