Caveman Shrink
CLI Verified ActiveMCP proxy that compresses prose fields (tool descriptions, etc.) using caveman rules. Same accuracy, fewer context tokens.
To reduce token consumption in AI agent workflows by compressing prose fields in MCP server responses, thereby lowering costs and improving efficiency.
Features
- Compresses prose fields in MCP messages
- Preserves code, URLs, paths, and identifiers
- Configurable fields to compress via environment variable
- Logs compression deltas when debug mode is enabled
Use Cases
- Reducing costs for AI agent workflows that consume large MCP tool catalogs
- Improving model inference speed by minimizing token count for tool descriptions
- Integrating with any MCP server to optimize its output for token efficiency
Non-Goals
- Compressing request bodies sent to the upstream server
- Mutating tool call responses (`tools/call`)
- Altering semantics or core functionality of tool descriptions
Code Execution
- info:LoggingDebug logging is available via `CAVEMAN_SHRINK_DEBUG`, but a persistent audit log for actions is not implemented.
Installation
npx -y caveman-shrinkQuality Score
VerifiedTrust Signals
Similar Extensions
Speclock Ros2
100AI Constraint Engine — enforces CLAUDE.md, .cursorrules, AGENTS.md rules as laws. 51 MCP tools, 991 tests. Official MCP Registry. npx speclock protect
Vercel CLI
99AI Elements is a component library and custom registry built on top of shadcn/ui to help you build AI-native applications faster.
Netdata Field Encoder CLI
99The fastest path to AI-powered full stack observability, even for lean teams.
Demo Metrics
99Open-source AI agent firewall for MCP security: agent egress control, DLP, SSRF, and prompt injection defense.
Gateguard
98A fact-forcing hook gate for Claude Code. Makes the AI pause and investigate before editing.