kdo
Workspace runtime for AI coding agents that caps context and serves it over MCP
Visit vivekpal1.github.io ↗What is kdo?
kdo is a workspace manager designed to help AI coding agents work efficiently in monorepos. Instead of agents traversing the entire filesystem and burning tokens on unnecessary files, kdo analyzes project structure, understands dependencies across Rust, TypeScript, Python, Anchor, and Go projects, and serves a budgeted view of the codebase via MCP protocol.
The tool enforces token budgets, detects agent loops (especially important for OpenClaw sessions), and works with Claude Code, OpenClaw, Cursor, and other MCP-compatible clients from a single endpoint. It reads build manifests like Cargo.toml and package.json, extracts function signatures with Tree-sitter, and keeps file bodies on disk until specifically requested, letting agents focus on what matters for their task.
Momentum
Daily activity over the last 30 days: upvotes, unique visits, new GitHub stars and Hacker News points.
How do we know it's vibe coded?
vibetrend discovered kdo automatically. Evidence: GitHub vivekpal1/kdo; repo has .cursorrules, AGENTS.md, CLAUDE.md. View source.
Similar vibe-coded apps
Open-source AI job search agent that runs locally in your CLI
Shared persistent memory for AI coding tools across sessions and applications
TypeScript framework for building AI agents without library dependencies
In-browser behavior tree execution engine for AI agents at ~60Hz
Local context layer that reduces AI coding agent token usage by 50-90%
