How deeply does AI Agents in Action, Second Edition cover context pruning?

Defines trimming through context summarization or removal and identifies a coherence trade-off, without prescribing a detailed workflow.

Context pruning

As explained in AI Agents in Action, Second Edition

Context trimming is the selective summarization or removal of conversation history, tool information, attachments, or other context before a model call.

Context trimming reduces the tokens sent on each model call by summarizing or removing conversation history, dropping irrelevant tool descriptions, or constraining response length with structured outputs. section 8.3.8

Unlike caching or model routing, which reduce costs through reuse or model selection, trimming changes the context sent to the model. Tight summaries can save tokens but weaken coherence across multi-turn conversations. section 8.3.8

Use trimming selectively:

Summarize or remove conversation history that does not need to be sent on every call. section 8.3.8
Drop irrelevant tool descriptions or use structured outputs to constrain response length. section 8.3.8
Measure cost per session, resolved task, and user before optimizing; weigh savings against coherence and the value being produced. section 8.3.8

How AI Agents in Action, Second Edition covers it

Cover of AI Agents in Action, Second Edition

liveBook manning.com

Read AI Agents in Action, Second Edition on liveBook Get the book at manning.com

Related concepts

ollama.chat()uses · ollama.chat() is a function in the Ollama Python library that sends conversational messages to a specified model and returns its response.
session historyuses · Session history is an ordered record of entries or interactions associated with a particular session.
pruningkind of · Pruning is the selective removal or restriction of parts of a model, structure, or candidate set, typically to reduce complexity or computation while retaining useful results.
costaddresses · Cost is the monetary expenditure, computational or other resource consumption, effort, or burden incurred in acquiring, producing, operating, or performing something.
structured datauses · Structured data is information organized according to an explicit schema or predictable fields, types, records, or relationships for consistent machine processing.

This explainer was generated by Manning's knowledge graph from the text of AI Agents in Action, Second Edition. The section links point to the parts of the book it draws on.