$ loading_
Compress and proxy MCP responses to reduce token usage for LLM tool calls.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "mcp-compressor" yet — see the docs or source repo.
Connect this MCP service to mcp-compressor, intercept all tool responses, compress the content to keep only information necessary for downstream reasoning, and report the token difference before and after compression.
Compressed tool responses plus a summary of token savings statistics.
Wrap the existing MCP toolset behind mcp-compressor, do not expose the raw tools directly, and provide only two meta-tools to the model with an explanation of the mapping.
A proxy setup exposing only two meta-tools, with a mapping from original tools to the meta-tools.
Analyze which responses in this MCP-based AI workflow consume the most context window space, and design a compression pipeline with mcp-compressor that prioritizes lower cost while preserving output quality.
Recommendations for a compression pipeline, response types suitable for compression, and cost-versus-quality tradeoff notes.
Compress MCP tool schemas to cut tokens while preserving semantics deterministically.
Proxy multiple MCP servers while reducing token usage with on-demand tool loading.
Compress prompts, tool outputs, and replies to reduce LLM token costs.
Aggregate, filter, transform, and compose MCP tools through one proxy.
Run a local-first MCP proxy with secure discovery and major token savings.
Aggregate multiple MCP servers into one for search, parallel calls, and orchestration.