@focusmcp/compress

Compress text output to save tokens — strip comments, collapse whitespace, remove blank lines, and abbreviate common patterns.

Measured economy

Static benchmark — no LLM, no variance, reproducible:

| Tool | Native equivalent | Brick output | Savings | |---|---:|---:|---:| | cmp_terse | 8,522 tokens | 233 tokens | -97.2% | | cmp_output | 8,522 tokens | 6,605 tokens | -19.9% |

Conditions: payload-only measurement (iso-call). Multi-turn agent dynamics not modeled. See global benchmark for methodology and full per-tool table.

Tools

| Tool | Exposed as | Description | |------|-----------|-------------| | output | cmp_output | Strip comments, collapse whitespace, remove blank lines (light/medium/aggressive) | | response | cmp_response | Compress JSON responses: strip nulls, shorten paths | | terse | cmp_terse | Ultra-compressed: keep only identifier names |