### Lossless JSON Token Reduction (`blackvall-token-compressor`)
Large JSON payloads (API responses, AST syntax trees, database records, and directory manifests) rapidly exhaust model context windows and cause massive token billing spikes.
`blackvall-token-compressor` strips repetitive JSON syntax overhead (quotes, braces, repeated keys across arrays of objects) before model ingestion while preserving 100% round-trip data fidelity.
#### Core Capabilities
- **35% to 60% Token Savings:** Proven reduction on structured arrays, tabular exports, and nested objects.
- **Lossless Two-Way Conversion:** Convert JSON into compact structural notation before model ingestion and restore deterministically back to valid JSON.
- **Dynamic Tool Schema Pruning:** Strip unused optional parameters and verbose schema descriptions from large tool catalogs before prompt injection.
- **Terminal Paste Buffer Interception:** Automatically compress large bracketed JSON pastes in terminal coding sessions.
Características Principales
01Cuts JSON context token usage by 35%–60% with zero data loss
02Works with Claude Code, Cursor, Windsurf, Cline & AGY out of the box
03Deterministic two-way JSON <-> compact structural conversion
04Prunes verbose unused parameters from large MCP tool schemas
05Intercepts and compresses large terminal bracketed pastes in flight
06100% local execution with zero external API keys required
Casos de Uso
01Compress large REST or GraphQL JSON responses before agent analysis
02Ingest multi-thousand-row database exports without hitting context limits
03Trim bloated OpenAPI and MCP tool schemas to save per-turn prompt tokens
04Pass AST syntax trees and compiler diagnostics between subagents compactly
05Benchmark and audit repository-wide JSON fixture token waste before CI runs
06Paste 50KB+ JSON logs into terminal coding CLIs without buffer truncation