Tag: agents
All posts tagged "agents".
How Do Real Needs Become a Direction for Agent Improvement?
8 min readTurn real work into reproducible, comparable tasks through domain modeling and evaluation, establishing conditions for later exploration and training.
Turning Occasional Success into Reliable Capability
8 min readFollow a configuration repair through exploration, success evaluation, demonstrations, rewards, and held-out testing to see how products close the learning loop.
Put What the Model Cannot Yet Do into the Harness
8 min readFollow the strengths and gaps of retrieval, memory, workflows, and tools to understand how Skills organize working methods, with an analogy to human memory.
Where Does a Harness Get Its Determinism?
9 min readDerive scheduling, isolation, authorization, injection defenses, recovery, and acceptance from the properties of agent execution, and locate the contracts a harness must enforce.
As Models Become More Capable, What Remains in the Harness?
7 min readSeparate runtime infrastructure from capability support, examine their shared boundaries, and understand how training changes their responsibilities.
When Language Models Learn the Physics of a Text World
8 min readOnce text can act on the world, context engineering represents goals, objects, states, actions, and feedback as a working environment the model can understand and use.
How Can a Prompt Elicit a Model's Capabilities?
9 min readExplain roles, examples, reasoning prompts, and reflection through conditional computation, then reassess prompting for reasoning models and agents.