News & guides
Tool Call Retries: Retry the Transport, Not the Judgement
Agents retry constantly, and most of it is wasted. How to tell a transport failure from a wrong decision, and what each one actually needs.
· 8 min readTool Result Formatting: Shaping What the Model Reads Back
Tool output is the largest block of prompt content you never wrote. How to shape it so agents stay accurate, cheap and able to decide what to do next.
· 8 min read
Tool Schema Design: The Prompt You Forgot You Were Writing
Tool definitions are prompt content the model reads on every turn. Design choices there change agent reliability more than model selection does.
· 8 min read
Tracing Multi-Step Agents: Building a Trace Tree That Holds
A flat log of model calls cannot explain a twenty-step run. How to shape spans, propagate context through subagents and async tools, and replay a run from its trace.
· 9 min read
Verifier Agents: Gating Completion on a Check That Passes
The most reliable agent improvement is refusing to accept completion until something verifies it. How to build a verifier that is worth trusting.
· 9 min read
Agent Cost Control: Patterns That Bound the Worst Case
Agent spend is not a per-call problem, it is a per-loop problem. Budgets, caching, tiering and circuit breakers that cap what a bad run can cost you.
· 9 min readAgent Error Recovery: Patterns That Keep the Loop Alive
Errors are the normal case in an agent loop, not the exception. These are the recovery patterns that separate agents that finish from agents that spiral.
· 9 min readAgent Memory: Managing Context Before It Manages You
Long agent sessions fail because context fills with noise. Here are the practical strategies for deciding what an agent should remember and what to drop.
· 8 min readAgent Observability: Tracing a Loop You Cannot Reproduce
Agent failures are rarely reproducible, so logs are not enough. What to record per step, how to span a tool loop, and which metrics predict a bad run.
· 9 min readAgent Planning Strategies: From ReAct to Explicit Plans
Reactive loops drift on long tasks and rigid plans break on contact with reality. Here are the planning strategies that work, and how to choose between them.
· 10 min readAgent Sandboxing: Containers, gVisor and microVMs
An agent that runs code needs a blast radius, not trust. How containers, gVisor and microVMs differ, and the network and credential controls that matter.
· 9 min read
AI vs AI Agent: What Actually Changes When You Add a Loop
A model answers. An agent decides, acts, and checks its work. The difference is a loop and a set of tools — here is exactly what that means in code.
· 8 min readReady to try it?
Every model behind one key. Start free and upgrade when it sticks.