Best Token AI Skills & MCP Servers

codesession — Track AI agent costs, tokens, files & commits. Session tracker for Claude Code, OpenClaw, Codex, GPT, Cursor, Windsurf, Cline & any AI agent. Budget enforcement, auto-pricing, MCP server, web dashboard. Also known as codesession-cli, code se

Cross-platform CLI tool for AI Game Developer (Skills & MCP). Full AI develop and test loop. Efficient token usage, advanced tools. Creates Unity project, installs plugins, configures tools, and manages HTTP connection with Unity Editor and a game made wi

Drop-in optimization toolkit for Claude Code — 15 agents, token savings, safety hooks, UI pipeline

MCP server for myOperator UI components - enables AI assistants to access component metadata, examples, and design tokens

Model Context Protocol server for swatchbook — exposes a DTCG project's tokens, axes, and diagnostics to AI agents without running Storybook.

MCP server for building Vue.js design systems - component analysis, documentation, variant generation, and design tokens

ctxloom — The Universal Code Context Engine. A local-first MCP server providing intelligent code context via hybrid Vector + AST + Graph search with Skeletonization (92% token reduction).

Shared MCP server building blocks for @us-all servers — token-efficient by design (category toggles, extractFields, search-tools meta, registry, runtime)

Focus your AI agents on what matters. 68+ bricks, 1 MCP server, modular context — from 200k to 2k tokens. Works with Claude Code, Cursor, Codex.

TES SDK — LLM observability and lifecycle tracking via Pentatonic Thing Event System. Track token usage, tool calls, and conversations. Manage things through event-sourced lifecycle stages with AI enrichment and vector search.

Model-agnostic code memory MCP server. Budget-aware graph retrieval for AI agents — sub-millisecond queries, token budgeting, no embeddings, no API keys. Built on CodeGraphContext for 155-language tree-sitter indexing.

Model Context Protocol server for Spectrum design data including tokens, schemas, and component anatomy

Local MCP server for PixelAgent — apply diffs, resolve elements, read design tokens

YouTube MCP server for token-optimized, structured data using the YouTube Data API v3.

Unified AI coding agent with MCP integration for Claude, VS Code, and Kiro IDE. Features persistent knowledge graph via OverdriveDB, token compression (60-80%), multi-model switching, and shared/independent project modes.

Token-efficient, markdown-first Notion MCP server — agents write markdown, never touch raw Notion JSON

Arguslog MCP server — exposes the full Arguslog API (issues, projects, orgs, alerts, releases, tier admin, …) as Model Context Protocol tools. Authenticate with a Personal Access Token from the dashboard.

Session metrics tracking — token usage, costs, duration, and per-tool statistics.

Token budget management — estimate and optimize token usage for AI agents.

Composite brick — orchestrates smartread, cache, compress, tokenbudget, and overview to deliver optimal context within a token budget.

Pack files into a compressed context bundle — reduce token usage by extracting only signatures, maps, or full content.

MCP server for BYTE Library — per-byte USDC data feeds for AI agents on Arbitrum. Subscribe to first-party feeds or pay-per-call via x402; every settlement carries an EIP-712 PayloadAttestation receipt. No API keys, no token.

JD Codec — local MCP proxy that compresses browser-agent snapshots ~80%+ via a cloud codec, with on-device PII redaction. The privacy-first connector for Playwright MCP and any MCP-capable agent.

Track and report token savings — compare FocusMCP usage vs naive full-read, trend over time, ROI.