L4 Acting in the world
Tool dispatch
Defines each tool once, checks every call before it runs, and hands every failure back to the model as a result it can fix.
Why it matters
Almost everything an agent does to the world passes through its tool calls. A call that runs unchecked, runs after being cut off, or crashes the turn turns a small model mistake into lost work or an unapproved action.
What to build
One typed definition, one validating gate
The default the research recommends. The reasoning and the code behind it are in the app.
What it takes, by kind of agent
- Conversational assistantMust-have
- CodingMust-have
- Deep researchMust-have
- Browser/GUIMust-have
How well it is studied
- Harnesses checked
- 30
- Checked facts behind it
- 397
See the 397 checked facts and how each harness does it
The facts, the code lines they come from and the rules for this topic are in the app. Sign in with Google; new accounts are approved by hand.
Also in L4 Acting in the world
- Perception and groundingDepends on agent type
Lets a browser or screen agent act on named elements, look again before every move, report missed actions plainly, and keep screenshots from flooding the context.
- Assistant Not applicable
- Coding Not applicable
- Research Not applicable
- Browser Must-have
- Extension hooksEmerging
Fixed points in the loop where plugins can watch, change or veto what the agent does.
- Assistant Recommended
- Coding Recommended
- Research Optional
- Browser Optional