New Workflows: Context Engineering¶
Why context quality drives output quality, and how to engineer context deliberately.
Why this chapter matters¶
Context quality determines output quality. AI systems perform best when repository conventions, tool access, and task intent are made explicit.
Key points for your team¶
When output is locally plausible but globally wrong, the root cause is often context quality rather than model quality. This chapter highlights context as an engineering asset that should be curated deliberately.
Attendees can apply this immediately by improving repository guidance and tool boundaries so generated changes reflect architectural intent, operational constraints, and domain standards.
What to review with your team¶
For team discussion, use this chapter to connect AGENTS.md - how this repo prefers to be edited, MCP servers - wiki, observability, internal APIs as tools, Per-task context packets for non-trivial work, and Bad output is usually context starvation with your current delivery loop.
In the session context, Context is engineered, not assumed. Three layers. Use that framing to align engineering, platform, and governance stakeholders on concrete next steps.
Put this into practice¶
Create or refine AGENTS.md and package context for complex tasks so generated changes reflect system-level intent, not just local code patterns.