compaction
github.com/philipppohlmann/compaction- Category
- Developer Tools
- Rank
- No. 1780Tools index
Previous survey · No. 1868 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Coding
- GitHub
- 6 stars
- Latest release
- v0.6.11
- Date
About
Compaction is a local tool that runs underneath Claude Code, Codex, and Cursor to shrink the input and output context sent to the model provider before it leaves the machine, without requiring a new editor or agent. Its own acceptance run reported an 8.39M-to-7.21M token input reduction (about 14%) and roughly 25% less generated output, with the measured before/after numbers shown inside the existing tools.
At a glance
Compaction sits between a coding tool and its model provider and reduces some of the text sent or requested. It is worth evaluating if repeated long sessions are costly and you can measure whether the agent still completes the same work.
Good fit
Developers running repeatable agent tasks who can compare token use and outcomes with a baseline.
Consider something else when
Keep your direct setup when policy prohibits local request interception or you need an independently verified quota forecast.
Start here
Run the same synthetic coding task with and without Compaction. Compare task success and measured input tokens, then check how to restore the original configuration.
Cost and requirements
Apache-2.0 project with documented support for Claude Code, Codex and Cursor. The publisher says eligible input reduction requires a free account.
Limits to know
The gateway can handle model-visible content, and the README says original requests may be retained locally for recovery. Estimated output savings and measured input counts are different evidence.
Sources checked 2026-11-05. Source-based assessment.
Intel on compaction
- Compact-Bench: Measuring What Context Compaction Costs Long-Horizon Coding Agents
- T3 Code Stable ships 1,400 changes, adds idle thread compaction
- Claude Code idle compaction targets cache-expiry costs
- Self-generated prompt injections in compaction summaries
- Lost in Compaction: Evaluating Side-Constraint Loss under Context Compaction
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.