Friction Lab / Products
Focused tools for the parts of developer workflow that generate the most friction. One released, several in progress.
Released
Wayfinder is the first public release from Friction Lab, available as a release candidate.
A Rust terminal navigator for macOS with shell-integrated navigation actions. Wayfinder lets you move through your filesystem, open files in your editor, copy paths, and reveal directories in Finder — all from a single keyboard-driven interface.
Designed for developers who spend significant time navigating large project trees and want a fast, scriptable alternative to file pickers and Finder.
In development
These tools are under active development and not yet available publicly.
A local-first macOS mission-control app for AI-assisted software development workflows. Relay surfaces context from your active project, routes tasks to AI agents, and keeps every step reviewable before it touches your codebase.
Read-only MCP server integration and Codex validation are already working in the current build. No public release is planned until the core workflow loop is stable.
Safe, explainable cleanup for generated developer artifacts — derived data, build caches, simulator runtimes, and other machine-generated directories that accumulate over time.
Cleanroom uses a profile-based cleanup model: you define what a clean state looks like, and the tool audits and reports before removing anything. No silent deletions.
Profile-based cleanup and audit-first safety model are in the design phase. Xcode-derived data and simulator runtime cleanup are the initial targets.
Roadmap
Ideas with clear problem statements that are queued for future development.
A repo-native smart documentation and memory layer for codebases and AI-assisted development. Code Atlas stores structured knowledge about your project alongside the code itself, making that knowledge accessible to both developers and AI agents via MCP.
A macOS workspace companion for task-focused desktops. Anchors specific apps and windows to named spaces so context-switching between projects is deterministic.
Finder folder icon suggestions using lightweight rule-based classification and optional ML-assisted inference. Keeps your project directories visually distinct without manual tagging.