This release is focused on one idea:
Give AI coding agents better context, while reducing the amount of unnecessary context they need.
When working across Claude Code, Cursor, Windsurf, Copilot, Cline, Antigravity, and other coding agents, a lot of tokens are wasted repeatedly discovering the same repository structure, reading the same files, and rebuilding context.
TokenCap is built to solve that.
What’s new in v2.1.0:
Code Simplification
tokencap analyze simplify
Find unreachable code, duplicate logic, unnecessary abstractions, and reinvention patterns across JS/TS with evidence and confidence levels.
Context Compression
tokencap compress
Compress CLAUDE.md, AGENTS.md, .cursorrules, and similar context files while preserving headings, code fences, links, and URLs.
Token Savings
tokencap stats
Track how much context and tokens you actually save across runs.
Debt Ledger
tokencap debt
Track TODOs, FIXMEs, and deferred shortcuts without relying on expensive git blame operations.
MCP v2
Expose repository overviews, symbol impact, call graphs, and delta diffs directly to MCP-compatible agents and IDEs.
Also crossed 2,500+ installs on the VS Code Marketplace, which is honestly pretty cool to see for an open source project I started building for my own workflow.
The idea behind TokenCap is simple:
Your AI coding agent should not have to rediscover your codebase every time you start a session.
Website:
https://tokencap.vansharora.app
VS Code Marketplace:
https://marketplace.visualstudio.com/items?itemName=tokencap.tokencap
GitHub:
https://github.com/vansharora21/tokencap-plugin
Install:
npx tokencap make