v1.9.0 proved that local-first repository intelligence can scale, crossing 2,000 downloads and establishing the performance foundation: bounded graph builds, sharded caching, workspace federation, and compression.
v2.0.0 builds on that foundation with a different goal: better understanding.
It adds:
Tree-sitter AST parsing across seven languages and symbol-level dependency graphs.
Full-stack frontend tracking for components, state, routes, accessibility, and design tokens.
Scalable, workspace-aware package context with sub-second incremental rebuilds.
Local-first privacy, secret redaction, and host-aware AI entry points
The important part is not generating more documentation. It is giving an AI coding agent the right repository context before it starts guessing.
A new R&D direction is also taking shape: TokenCap Companion.
The idea is to bridge browser AI chats and local project intelligence in both directions, without relying on a hosted server, LLM API credits, or uploading source code. It is early research, but the objective is clear: make useful context move between local code and web chat while keeping control with the developer.
TokenCap v2.0.0 is a meaningful step toward that.
Website
https://lnkd.in/gYxw5Mum
Documentation
https://lnkd.in/grNwdY-P
GitHub
https://lnkd.in/g5rvbyCm
npm
https://lnkd.in/gyZsGXnr
VS Code Marketplace
https://lnkd.in/gCJmJdN6
Open VSX
https://lnkd.in/gF2rT_aY
Product Hunt :
https://lnkd.in/ga9vyM5n