Get your free personalized podcast brief
We scan new podcasts and send you the top 5 insights daily.
Instead of loading large context files on every turn, use "skills." The agent only sees a skill's name and description initially, loading the full instructions only when needed. This method, called progressive disclosure, drastically saves tokens and improves performance.
Related Insights
Modern AI models infer context from the codebase, making detailed `agent.md` files redundant. These files waste tokens on every interaction and are only necessary for highly specific, proprietary information that must always be present in the context.
Counterintuitively, the goal of Claude's `.clodmd` files is not to load maximum data, but to create lean indexes. This guides the AI agent to load only the most relevant context for a query, preserving its limited "thinking room" and preventing overload.
Instead of a single, monolithic "About Me" file, structure personal context into modular files (e.g., roles, projects, team). This design allows you to provide an AI agent with only the specific information it needs for a given task, which enhances efficiency, relevance, and privacy.
Don't write agent skills from scratch. First, manually guide the agent through a workflow step-by-step. After a successful run, instruct the agent to review that conversation history and generate the skill from it. This provides the crucial context of what a successful outcome looks like.
The "Agent Skills" format was created by Anthropic to solve a key performance bottleneck. As capabilities were added, system prompts became too large, degrading speed and reliability. Skills use "progressive disclosure," loading only relevant information as needed, which preserves the context window for the task at hand.
Instead of overloading the context window, encapsulate deep domain knowledge into "skill" files. Claude Code can then intelligently pull in this information "just-in-time" when it needs to perform a specific task, like following a complex architectural pattern.
Simply giving an AI agent thousands of tools is counterproductive. The real value lies in an 'agentic tool execution layer' that provides just-in-time discovery and managed execution to prevent the agent from getting overwhelmed by its options.
Overloading LLMs with excessive context degrades performance, a phenomenon known as 'context rot'. Claude Skills address this by loading context only when relevant to a specific task. This laser-focused approach improves accuracy and avoids the performance degradation seen in broader project-level contexts.
Agent Skills only load a skill's full instructions after user confirmation. This multi-phase flow avoids bloating the context window with unused tools, saving on token costs and improving performance compared to a single large system prompt.
To make agents useful over long periods, Tasklet engineers an "illusion" of infinite memory. Instead of feeding a long chat history, they use advanced context engineering: LLM-based compaction, scoping context for sub-agents, and having the LLM manage its own state in a SQL database to recall relevant information efficiently.
