ai / agents
An agent is a software program that runs tools in a loop to achieve a goal.
You state the goal. The agent runs tools (web search, reading and writing files, etc.), observes results, and works until the goal is achieved. I run agents in Warp as my primary tool for software development.
Analogous terminology
| Horse-drawn freight | Electric truck | AI agent stack | |
|---|---|---|---|
| Operator | Driver | Driver | Human operator |
| Control | Reins | Steering wheel & pedals | Prompts & goal |
| Power | Horse | Battery & motor | Model |
| Harness | Leather harness | Wiring harness | Agent harness |
| Tools | Traces, axles & wheels | Shafts, axles & wheels | Tools & APIs |
| Load | Cargo payload | Cargo payload | The task |
A native terminal is the harness
I don't use AI IDEs like Cursor or Zed; I'm a long-time Vim user. I also find raw CLIs like Claude Code or Codex too limiting. The terminal is the harness, and Warp is a native UI built for it.
Some things are nicer with a native UI:
- Reviewing diffs inline to accept or reject hunks.
- Tracking the agent's active task list.
- Interleaving prompts and normal shell commands.
- Dragging screenshots or speaking prompts via voice.
Frictionless maintenance
Modern agents handle the mechanical work: features, refactoring, security patches, and bugs. They diagnose and fix error backtraces.
Our Sentry backlog has been at zero for most of 2026. Bugs and technical debt are addressed faster; maintenance is cheap, so there is less excuse to let things rot.
Humans on the loop
We are often humans in the loop, writing instructions and reviewing every change.
We are also humans on the loop: we design the harness itself. CI and tests are constraints we tighten around the agent.
Specifically, checks run more often locally and in CI. cibot starts CI runs fast and outputs plain text to paste back to the agent if a check fails.
We also check AGENTS.md files into the repo. They are version-controlled and
scoped by directory. A root AGENTS.md covers architecture and quick reference.
Subdirectory files (db/AGENTS.md, ui/AGENTS.md) have domain-specific rules.
What goes in AGENTS.md
Patterns I've found useful:
Quick reference. Runnable commands for common tasks like build, test, lint, and migrate.
Safety rails. Explicit rules like "never commit secrets."
Conditional checks. "Run X if Y files touched" to keep feedback fast.
Style guides. Commit message conventions and naming patterns.
Environment context. Available tools, port numbers, and file paths.
What comes next?
Today, I often run multiple agents in separate Warp tabs, inside git worktrees.
The next step might be orchestrating parallel agents in the cloud. Project boards like Linear track what we need to build; platforms like Warp Oz run agents in the cloud. If you connect the two, the board becomes the team's control plane.