Built with Claude · v1.0.0

token-saver

An MCP plugin that watches every AI output for token waste — fires warnings, errors, and alerts before your context burns.

🔍

Noise Detection

Detects logs, stack traces, ANSI codes, and repetitive history eating your context window.

🚨

Alert Levels

Four thresholds — info, warning, error, alert — configurable per project via .token-saver.json.

🌐

Model Agnostic

Works with Claude, GPT, Gemini, or any model output. Any MCP client. Zero lock-in.

📦

Zero Config

One npx -y token-saver-mcp and you're live. No build step needed.

token-saver-mcp · monitor mode · session active
info
Output analyzed · 312 tokens · type: tool_result
below warning threshold (1000) · no suppression
warning
Large log output detected · 4,218 tokens · 94% noise patterns
matches [INFO] [DEBUG] ANSI timestamps · suppressLogs: true → shouldSuppress
error
Repetitive history detected · 6,104 tokens · 3 near-duplicate turns
cumulative session waste: ~18,400 tokens · consider truncating history
Usage Details
Model claude-sonnet-4-6
Tokens used ~460,000 input · ~72,000 output  ~532k total
Session duration ~3 hours (2 sessions, context compacted once)
Tech stack TypeScript · Node.js · MCP SDK · GitHub Actions · semantic-release · Docker · GitHub Pages
What was achieved Designed, built, and shipped token-saver MCP v1.0.0 — full plugin with alert engine, session stats, suppress logic, website, CI/CD, npm publish, and Docker Hub release.
Estimated cost ~$2.46 USD  ($3/MTok in · $15/MTok out)