Compact
Conversation compression
Compact compresses your conversation history while preserving everything that matters — decisions, insights, file references, and outstanding questions. Fresh context window, zero lost knowledge.
How it works
Fresh context window,
zero lost knowledge
When your conversation gets long, run /compact. It summarizes the full history into a compressed form — keeping decisions, insights, and context while freeing the context window for new work.
Smart Compression
Keeps decisions, insights, file references, and outstanding questions. Discards implementation noise, tool call details, and repetitive exchanges.
Context Preservation
The AI continues with full knowledge of what came before. No re-explaining. No lost ground. The conversation picks up where you left off.
Focus Chain
Multiple compactions build a focus chain — a layered history of your thinking. Earlier compactions are preserved, creating a trail of how the work evolved.
Instant & Local
Runs entirely in the browser — no API call needed. Your compressed context is available immediately. No waiting, no cost.
Compress, not lose
Not "start over from scratch."
Keep everything that matters. Lose everything that doesn't.
When conversations get long
Long conversations. Short context. No loss.
Any session where the conversation outgrows the context window.
Deep Research Sessions
3-hour research dive → compressed to key findings and next steps
Multi-Day Projects
Days of context → distilled plan, decisions, and current state
Complex Debugging
Long debugging trail → what worked, what didn't, what's left
Strategy Discussions
Extended strategy session → decisions made and open questions
Collaborative Sessions
Multi-person discussions → shared context everyone can continue from
Writing Sessions
Long drafting session → key decisions about voice, structure, themes