Token Pilot
How It Works
Traditional: Read("user-service.ts") → 500 lines → ~3000 tokens Token Pilot: smart_read("user-service.ts") → 15-line outline → ~200 tokens
Up to 90% reduction on large files. Files under 200 lines returned in full (zero overhead for small files).
Benchmarks (real data)
Measured on public open-source repos (express, fastify, flask). Files ≥50 lines:
- token-pilot (TS): 48 files, 91% savings
- express (JS): 6 files, 99% savings
- fastify (JS): 23 files, 94% savings
- flask (Python): 20 files, 91% savings
- Total: 97 files, 231K → 19K tokens (92% savings)
Quick Start
npx -y token-pilot init
Creates .mcp.json automatically. Restart your AI assistant — done.
18 MCP Tools
Core Reading: smart_read, read_symbol, read_for_edit, read_range, read_diff, smart_read_many Search: find_usages, project_overview, related_files, outline, find_unused, code_audit, module_info Git: smart_diff, explore_area, smart_log, test_summary Analytics: session_analytics
29 Languages
TypeScript, JavaScript, Python, Rust, Go, Java, Kotlin, Swift, C#, C++, C, PHP, Ruby, Scala, Dart, Lua, Shell, SQL, R, Vue, Svelte, Perl, Groovy, Elixir, Common Lisp, Matlab, Protocol Buffers, and more.
Server Config
{
"mcpServers": {
"token-pilot": {
"command": "npx",
"args": [
"-y",
"token-pilot"
]
}
}
}