Smart terminal filters
Built-in, flag-aware filters cut repetitive terminal output while preserving failures, summaries, and the clues your agent needs next.
How token savings work
boost features
Less noise. More capable agents.
Explore how Boost saves context, explains agent work, and fits the developer tools your team already uses.
Custom filters
Create TOML filters for your internal CLIs and scripts, or let your coding agent mine local Boost history and draft the highest-impact filters for you.
Create a custom filter
MCP filters
Staged rolloutCompress large MCP JSON responses before they enter agent context. Boost keeps a local record of compressed calls and the context saved.
Configure MCP compression
Sharper codebase exploration
Early accessBoostGraph gives agents structural code search instead of long search-and-read loops. In early benchmarks, exploration completed 3× faster and cost up to 50% less.
BoostGraph feature guideEvery session, explained
Open a session to see its models, subagents, turns, tool calls, token usage, and the filters that saved context. Provider counts and estimates stay clearly separated.
Why Boost measures locally
Navigate all your sessions
Move between parent sessions and subagents, inspect the tools each one used, and open before-and-after filter diffs without leaving the report.
GitHub PR usage comments
Staged rolloutKeep one live pull-request comment with token usage and estimated cost broken down by coding-agent harness and model.
Set up PR usage commentsFlexible by default
Boost works out of the box, then adapts to how your team builds. Shape filtering, reporting, privacy, and updates without changing your agents or everyday commands.
Explore configuration