In early 2025, LinkedIn tackled a common issue with AI coding agents: their lack of understanding of organizational context. While these agents excel at coding tasks, they struggle with specifics related to LinkedIn's frameworks, services, and internal systems. New engineers spend months learning the ropes, and even experienced ones hit walls when facing unfamiliar components. To address this, LinkedIn developed the Contextual Agent Playbooks & Tools (CAPT), which combine deep organizational knowledge with executable workflows to enhance AI coding agents' effectiveness. CAPT is built on the Model Context Protocol (MCP), allowing AI agents to access LinkedIn’s internal systems and third-party services. This integration provides standardized interfaces for various tools and supports community-built resources. The real power of CAPT lies in its playbooks, which transform institutional knowledge into step-by-step workflows. For instance, the experiment cleanup playbook streamlines the process of managing A/B tests, enabling any engineer to perform tasks that previously required specialized knowledge. This approach not only saves time but also democratizes access to expertise within the engineering teams. To ensure that CAPT is easy to adopt, the development focused on zero-friction distribution. CAPT is delivered as a Python package, automatically updating on engineers' laptops without complicated setup. Authentication with external systems is simplified to a single command or handled automatically when needed. This design promotes rapid adoption across the organization, minimizing barriers to entry. As a result, CAPT has fundamentally changed how LinkedIn engineers build, debug, and operate software, making AI assistance a more integral part of their workflow.