The software around language models that turns them into agents: loops, tools, context, permissions and evaluation.
25 in this topic
The software around a language model that turns it into an agent: the loop, tools, context, permissions and memory.
Agent harnesses 2 min read 27 Sep 2025
The core cycle every agent runs: think, call a tool, observe the result, repeat — and how the loop knows when to stop.
Agent harnesses 1 min read 26 Sep 2025
How to write tools agents use well: clear names, precise descriptions, sensible inputs and informative outputs.
Agent harnesses 1 min read 25 Sep 2025
How agents stay effective over long tasks: what to keep in context, what to summarise, and what to store outside.
Agent harnesses 1 min read 24 Sep 2025
Controlling what agents can do on their own: allow lists, approval prompts, forbidden actions and escalation.
Agent harnesses 1 min read 23 Sep 2025
Isolating agents so their mistakes and any manipulation can't reach sensitive systems: containers, VMs, file and network limits.
Agent harnesses 1 min read 22 Sep 2025
When and how an agent should hand work to other agents with their own context, and the trade-offs involved.
Agent harnesses 1 min read 21 Sep 2025
How agents remember across steps and sessions: working memory, notes files, retrieval and user preferences.
Agent harnesses 1 min read 20 Sep 2025
How to test agents that take many steps: task suites, success criteria, trajectory review and regression testing.
Agent harnesses 1 min read 19 Sep 2025
The tooling that runs models against benchmarks and test sets: datasets, prompting, scoring and reporting.
Agent harnesses 1 min read 18 Sep 2025
How agent harnesses cope when tools fail, models misbehave or tasks go wrong — retries, feedback and graceful stopping.
Agent harnesses 1 min read 17 Sep 2025
Logging and tracing agents' steps, tool calls, costs and decisions so you can debug, audit and improve them.
Agent harnesses 1 min read 16 Sep 2025
Agents that operate graphical interfaces by viewing screenshots and controlling the mouse and keyboard: uses and precautions.
Agent harnesses 1 min read 15 Sep 2025
How coding agents work: reading code, editing files, running tests and commands, and the safeguards that make them useful.
Agent harnesses 1 min read 14 Sep 2025
Agents that search, read and synthesise information from many sources, and how to keep their reports accurate.
Agent harnesses 1 min read 13 Sep 2025
Choosing between building your own agent loop and using an SDK or framework, and what to look for.
Agent harnesses 1 min read 12 Sep 2025
Using hooks to run your own code at points in the agent loop: validating actions, formatting code and enforcing policy.
Agent harnesses 1 min read 11 Sep 2025
How agents break down complex tasks, track progress and adapt plans as they learn more.
Agent harnesses 1 min read 10 Sep 2025
Designing agents that work with people: checkpoints, clarifying questions, reviewable output and handing back control.
Agent harnesses 1 min read 9 Sep 2025
Why agents can be expensive and slow, and how harness design keeps token usage and running time under control.
Agent harnesses 1 min read 8 Sep 2025