Small, sharp tools and conventions for working alongside AI coding agents — built around one idea: candor.
When you hand work to a coding agent, a lot happens that you can't see. How much it's costing you. How close you are to a usage wall. How much context is left before it silently forgets. Which model is actually answering. How hard it's thinking. The default experience hides all of it — so you find out you hit a limit when you get cut off, find out a session was expensive when you check a dashboard later, and find out context compacted when the agent suddenly loses the thread.
This toolkit's north star is candor — frankness, openness, honesty:
- Transparency — surface the state the agent normally hides, in the place you're already looking. Make the invisible legible.
- Openness — open source, and open about what the agent is doing on your behalf and what it costs.
- Empathy — respect the person doing the work with the truth, and lower their cognitive load instead of adding to it.
Transparency isn't a feature bolted on the side; for agentic coding it's the whole point. You can only steer what you can see. A tool that tells you the truth about your session — plainly, continuously, without you asking — is a tool that lets you make better calls: when to push, when to switch models, when to clear context, when to slow down. These are the same values we want from the agent itself; tooling that's candid about what's happening trains the habit of expecting candor everywhere.
| Module | What it is | Targets |
|---|---|---|
| code-status-bar | A status line that shows usage limits, cost, context health, and git/worktree state at a glance | Claude Code (via ccstatusline) |
| shift | An autonomous work-queue runner: pre-load bins of work, leave, and it keeps the agent grinding through them — past natural stop points and across rate-limit resets — leaving every decision logged and every change a reviewable commit | Claude Code (Stop hook + headless -p) |
New here? Start with the Code Status Bar. It installs as a portable, zero-dependency default, or an opt-in colored variant that recolors the usage bars green → yellow → red as you approach each limit — so you feel a wall coming before you read a single number. You could build it by hand in ccstatusline's editor; this is that setup already done — one command, no configuration, and still fully editable.
Going heads-down? shift turns an unattended run — the least transparent mode there is — into an honest paper trail: you trade real-time steering for a
shift/<date>branch, a decision log, and a "here's what I did and what needs you" summary. One command wires the hook; the safety model keeps the work on a branch and off your remotes.
More to come. Each module is self-contained, declares which agent it targets, and explains why every piece earns its place — because justifying the real estate is part of the philosophy.
MIT — see LICENSE.