switchboard
C

Claw Compactor

by OpenCompress

Compresses LLM context such as code, JSON, logs, and chat history through a reversible 14-stage pipeline with no LLM calls, via a Python API and CLI, for agent developers.

4
Skills
None
Auth
No
Streaming
No
Push

Skills

Fusion Pipeline Compression

Routes code, JSON, logs, diffs, and search results through content-aware stages to cut token counts without LLM inference.

Reversible Rewind Retrieval

Stores originals behind hash markers so the LLM can call a tool to retrieve any compressed section.

Chat Message Compression

Compresses chat message lists with cross-message deduplication before the per-message pipeline.

Workspace Commands

Benchmarks, compresses, deduplicates, and estimates tokens across a workspace, with a watch mode for file changes.

Infrastructure & OpsMemory & Statetoken-compressioncontext-windowreversible-compressionprompt-compressionpythoncost-reduction
Visit Agent
claw-compactor
Compresses LLM context such as code, JSON, logs, and chat history through a reversible 14-stage pipeline with no LLM calls, via a Python API and CLI, for agent developers.
fields
nameClaw Compactor
providerOpenCompress
urlhttps://github.com/open-compress/claw-compactor
categoriesinfrastructure · memory-state
accessapi · cli
authnone
streamingfalse
pushfalse
verifiedtrue
tagstoken-compression, context-window, reversible-compression, prompt-compression, python, cost-reduction
skills
fusion-compressionFusion Pipeline CompressionRoutes code, JSON, logs, diffs, and search results thro…
rewind-retrievalReversible Rewind RetrievalStores originals behind hash markers so the LLM can cal…
chat-compressionChat Message CompressionCompresses chat message lists with cross-message dedupl…
workspace-commandsWorkspace CommandsBenchmarks, compresses, deduplicates, and estimates tok…