treecity labs
the missing toolsfor AI coding agents
Talk to your agents instead of typing, capture context without breaking flow, give every project the right skills, and run it all from one workspace.
> cat ~/why-were-building-this.md
AI coding works. The workflow around it still feels stitched together.
We use coding agents every day, and the same friction keeps showing up. We looked for better tools around the agents we already use, never found exactly what we wanted, and started building our own.
> too much typing
Long prompts, code reviews, and explanations still get typed out one line at a time.
> context is scattered
Screenshots, browser tabs, files, logs, and errors rarely reach the agent cleanly.
> setup repeats itself
Skills, MCP servers, rules, and hooks get copied project to project, whether they belong there or not.
> ls ~/treecity
Four tools for working with AI coding agents
Each app solves one specific problem. They work on their own, and fit together when you want them to.
> treecity explain
Talk. Capture. Configure. Build.
talk
TreeSpeak
Tell the agent what you want without typing a wall of text.
capture
TreeSnap
Add the screenshot, error, or file it needs.
configure
TreeSkill
Load the right skills, tools, and rules.
build
TreeSpace
Run the project, terminals, and agents from one workspace.
You don't need all four. The point is they fit together when you want them to.
> ./watch-us-build
Watch us build these, in the open
New videos on the channel: building the tools, testing them, breaking them, and changing direction when something isn't working. Real dev logs and late-night sessions, not a highlight reel.