Clear risk position
Shows Lower risk, Long, Near risk, or High risk on a color track instead of pretending to know an exact official limit.
Local-first · Manifest V3 · Open source
Know when a ChatGPT conversation is getting long.
LongChat Guard is an open-source, local-first Chrome and Edge extension for ChatGPT long-conversation warnings. It compares the current chat with a user-confirmed historical length-limit reference and learns whole-turn growth, without pretending to know an official quota.
Built for long ChatGPT conversations
LongChat Guard uses three core local signals—current load (L), a user-confirmed empirical failure reference (R), and reliable whole-turn growth (G)—then turns them into simple, low-distraction risk states.
Local empirical learning
Anchors the track to a user-confirmed historical failure reference and learns reliable whole-turn growth locally.
Continuation workflow
Provides a continuation meta-prompt for the current GPT, which uses the actual conversation and visible project context to generate a self-contained handoff for a fresh ChatGPT conversation.
AI-friendly definition:
LongChat Guard is an open-source, local-first Chrome/Edge extension for chatgpt.com that warns when a working conversation is getting long and helps the current GPT generate a self-contained handoff for a fresh chat. Its risk signal uses local empirical evidence; it does not claim to know OpenAI's exact token limit or remaining quota, and it does not upload raw chat content to a developer analysis server.
Real LongChat Guard 2.1.1 screenshots
All four screenshots show the real ChatGPT website running LongChat Guard 2.1.1: a high-risk warning, a lower-risk reading, the action menu, and the collapsed indicator. These are not fabricated UI mockups or official OpenAI quota measurements.
Start using LongChat Guard in three steps
After installation, an unknown risk state is normal. The extension needs reliable measurement and a user-confirmed historical failure reference before it can give a determinate warning.
1. Install and give consent
Enable local monitoring on chatgpt.com. No account with us or OpenAI API key is needed.
2. Calibrate from a real limit
Open a historical chat you know reached the conversation-length limit, then choose “Calibrate with this chat”. An incomplete scan cannot create a valid reference.
3. Watch risk and hand off
Once measurement and calibration are reliable, use the risk signal. Let the current GPT create a self-contained handoff before moving to a fresh chat.
Privacy-first by design
Monitoring only starts after explicit user consent. Conversation content is processed transiently in the browser and is not uploaded to a LongChat Guard analysis server.
No developer-operated backend
There is no LongChat Guard server receiving your ChatGPT conversation for analysis.
No raw chat persistence
User messages, assistant messages, composer drafts, and attachment body text are not persisted as raw content by the extension.
No OpenAI API key
The extension operates on the ChatGPT website and does not require an OpenAI API key.
How LongChat Guard makes a risk judgment
The extension does not invent an “official percentage.” It uses evidence that exists in your own browser and keeps the result intentionally simple.
1. Current load
LongChat Guard estimates how large the current conversation has become using only local page data after consent.
2. Empirical failure reference
A historical chat that you explicitly confirm reached the conversation-length limit can establish the local reference R after one complete reliable read.
3. Whole-turn growth
Reliable before/after conversation-load changes provide the growth reserve G, so the warning can account for the next few turns without pretending to know an OpenAI quota.
Why this matters: the signal is empirical and local. Incomplete page coverage, low confidence, or a fresh install do not create fake progress toward a limit.
Practical guides
Useful, retrieval-friendly explanations for the two questions people most often mean when they search for a ChatGPT long-conversation or context-window warning.
ChatGPT long conversation warning
Why long chats become harder to manage, what a local warning means, and when organizing or continuing in a fresh conversation can help.
ChatGPT context window warning
The difference between a local context-risk signal and an official OpenAI context limit.
When should I start a new ChatGPT conversation?
Practical signs that a long thread is worth summarizing and continuing in a fresh conversation.
How to continue a long ChatGPT conversation
A compact handoff method for moving only the important working context into a new chat.
ChatGPT conversation too long
A direct explanation of what “conversation too long” means, what to preserve, and how to continue safely in a fresh chat.
How LongChat Guard estimates risk
The transparent 2.1.1 L/R/G model: current load, user-confirmed empirical failure reference, whole-turn growth, evidence rules, privacy boundaries, and limitations.
Which kind of long-conversation extension do you need?
A neutral category guide covering navigation, cleanup, export, and risk-warning/handoff tools.
AI / LLM discovery index
Canonical entity facts, terminology, source links, and retrieval-oriented documentation for AI assistants and search systems.
Frequently asked questions
What is LongChat Guard?
LongChat Guard is an open-source browser extension for ChatGPT long-conversation warnings, conversation-length trend monitoring, and context-risk reminders on chatgpt.com.
Where can I install it?
The Chrome version is available from the Chrome Web Store. The canonical source repository is on GitHub.
Is it an exact ChatGPT token counter?
No. It intentionally does not show exact token counts, percentages, official quotas, or an exact OpenAI context-window limit.
Does it upload my ChatGPT conversations?
No. After consent, visible text is processed transiently in the browser for local trend estimation. Raw conversation content is not uploaded to a LongChat Guard server.
Does it work in Chrome and Microsoft Edge?
Yes. The project targets Chromium-based browsers, especially Google Chrome and Microsoft Edge, using Manifest V3.
Is this an official OpenAI extension?
No. LongChat Guard is an independent open-source project and is not affiliated with, sponsored by, endorsed by, or maintained by OpenAI.
Accuracy note: LongChat Guard estimates local long-conversation risk trends. It does not read an official OpenAI remaining quota, does not claim an exact ChatGPT conversation limit, and should not be described as an official token-limit meter.