Frequently asked questions
Does ChatCode upload my entire source code to the server?
ChatCode does not bulk-upload your repository to its server for storage. The project remains on the computer running ChatCode. However, when AI requests reading/analysis, the relevant context may be sent to the AI provider through the conversation. See Data & security.
Does a Project have to be source code?
No. A Project is simply a folder where you let AI work: code, Markdown/TXT/CSV/JSON, a knowledge base, or an empty Output folder.
Can I add an entire D: drive?
Yes. For a broad scope, keep it read-only and separate a writable Output project. See example.
Does a multi-GB project get dumped into the AI context?
No. Project Brain helps locate relevant files/symbols/context so the AI can retrieve what it needs instead of stuffing the entire workspace into one prompt.
Why does ChatGPT need a URL?
The URL lets ChatGPT reach the correct ChatCode instance without requiring you to expose your IP or configure your router. See the detailed explanation.
Free and URL cố định khác nhau thế nào?
If the effective policy does not grant a fixed URL, ChatCode uses a temporary link that may change. When fixed-URL entitlement is enabled for the plan, account, or license, the connection address remains stable. View the guide.
What if my connection link leaks?
Open ChatCode, bấm Generate new code, sau đó bấm Copy connection link and update the connector in your AI. Links using the old code will lose access; you do not need to manually change the URL. In Advanced Security mode, the AI also needs valid OAuth authorization. See recovery steps.
Is Advanced Security mandatory?
No. You can use Standard mode for quick setup or enable Advanced Security to require OAuth. For sensitive repositories or company computers, Advanced Security is recommended. Compare both modes.
What happens if my computer is off?
ChatGPT cannot operate on a project while the computer holding it is off. With a fixed URL, the address remains yours, but ChatGPT will report the device as disconnected until ChatCode comes online again.
Can I assign work from my phone?
Yes, if your AI platform supports custom connectors on mobile. For example, Claude can use remote MCP servers previously added via claude.ai, while ChatGPT custom MCP apps are currently web-only. See AI compatibility matrix.
What plans does ChatCode currently offer?
During Beta there are only Free $0 and Pro Beta $12/month. Free has usage limits without a published fixed numeric quota. Pro Beta unlocks every Beta feature with unlimited normal use. Purchase and renewal are monthly only; there is no Lite, annual, or Lifetime plan yet. Commercial-release packaging and pricing may change after Beta.
Does ChatCode make ChatGPT unlimited?
No. Pro Beta's “unlimited” applies only to ChatCode usage under fair-use, security, and anti-abuse safeguards. Limits, models, features, and policies for ChatGPT, Claude, Grok, or Mistral are still controlled by the AI provider.
How does ChatCode calculate tokens and savings?
ChatCode measures only data that passes through ChatCode: AI → ChatCode tool-call requests and ChatCode → AI results/code/context. Provider chat history, Memory, system prompts, reasoning, and normal chat responses are excluded. Tokens are estimated from ChatCode payloads; for Smart Brain tasks, ChatCode compares the actual context sent with a broad project-context reference to estimate reduction. If there is not enough data, ChatCode does not invent a savings percentage.
What usage data does OzinTech receive?
When aggregate sharing is enabled, the server receives only provider/reference-model, counters, estimated bytes/tokens, coarse project-size bucket, app version/plan, and operation category. It does not receive prompts, source code, file names or paths, chat history, connector URLs, tokens, or secrets. Aggregates help improve Project Brain, evaluate Beta usage, and allocate resources appropriately.