Feature summary
Add context usage and limits indicator to each session
What problem are you trying to solve?
Problem
When using GitHub Desktop’s built-in chat extensively, it’s difficult to understand how much of the model’s available context window has been consumed.
As conversations grow, responses may become slower, older context may be truncated, or the model may lose awareness of earlier parts of the discussion. Currently there is no visible indication of:
- Current context usage
- Remaining context capacity
- When older messages are likely to be dropped
- Whether attached files, repository context, and chat history are contributing significantly to context consumption
This makes it difficult to manage long-running technical sessions, architecture discussions, debugging workflows, and code reviews.
Alternatives considered
Users can currently estimate context usage indirectly based on conversation length, but this is imprecise and varies significantly depending on attached files, repository size, and model selection.
Proposed solution
Add a context usage indicator to the chat interface.
Possible implementations:
- Simple percentage indicator (e.g. “62% of context used”)
- Token count display (e.g. “145k / 256k tokens”)
- Progress bar in the chat header
- Expandable details panel showing context contributors:
- Chat messages
- Repository context
- Attached files
- MCP/tool responses
- Other injected context
Workflow impact
- Helps users understand when a conversation is approaching context limits.
- Encourages proactive summarization before important context is lost.
- Improves trust and predictability during long technical sessions.
- Makes performance and response quality easier to reason about.
- Particularly valuable for power users working on large repositories and multi-hour development workflows.
Installation context
No response
Additional context
Many AI-powered development tools expose some form of context-window visibility. GitHub Desktop already provides a sophisticated chat experience, but the absence of context utilization metrics makes it difficult to manage longer sessions effectively.
Even an approximate value would be more useful than no visibility at all.
Feature summary
Add context usage and limits indicator to each session
What problem are you trying to solve?
Problem
When using GitHub Desktop’s built-in chat extensively, it’s difficult to understand how much of the model’s available context window has been consumed.
As conversations grow, responses may become slower, older context may be truncated, or the model may lose awareness of earlier parts of the discussion. Currently there is no visible indication of:
This makes it difficult to manage long-running technical sessions, architecture discussions, debugging workflows, and code reviews.
Alternatives considered
Users can currently estimate context usage indirectly based on conversation length, but this is imprecise and varies significantly depending on attached files, repository size, and model selection.
Proposed solution
Add a context usage indicator to the chat interface.
Possible implementations:
Workflow impact
Installation context
No response
Additional context
Many AI-powered development tools expose some form of context-window visibility. GitHub Desktop already provides a sophisticated chat experience, but the absence of context utilization metrics makes it difficult to manage longer sessions effectively.
Even an approximate value would be more useful than no visibility at all.