February 22, 2026
An agent with access to 30 tools is not automatically more capable than one with 5. What matters is how it selects, sequences, and validates tool use. Poor tool use wastes tokens, introduces latency, and produces wrong results that look right.
When multiple tools could handle a task, the agent must pick the best one. This is harder than it sounds because tool descriptions are imperfect and tasks are ambiguous.
February 22, 2026
A 32B model with a 32K context window can process roughly 8-10 source files at once. A real codebase has hundreds. Concatenating everything into one prompt fails — the context overflows, quality degrades, and the model either truncates or hallucinates connections.
The two-pass pattern solves this by splitting analysis into two stages:
This effectively multiplies your context window by the compression ratio of summarization — typically 10-20x. A 32K context that handles 10 files directly can handle 100-200 files through summaries.
February 22, 2026
Not every infrastructure change needs a full Kubernetes cluster to validate. Some changes can be verified with a linter in under a second. Others genuinely need a multi-node cluster with ingress, persistent volumes, and network policies. The cost of choosing wrong is real in both directions: too little validation lets broken configs reach production, while too much wastes minutes or hours on environments you did not need.
February 22, 2026
A validation playbook is a structured procedure that tells an agent exactly how to validate a specific type of infrastructure change. The key problem it solves: the same validation (for example, “verify this Helm chart works”) requires different commands depending on whether the agent has access to kind, minikube, a cloud cluster, or nothing but a linter. A playbook encodes all path variants in one document so the agent picks the right commands for its environment.
February 21, 2026
Agents are stateless by default. Every new session starts with a blank slate – no knowledge of previous conversations, past mistakes, or learned preferences. This is the fundamental problem of agent context management: how do you give an agent continuity without overwhelming its context window?
Agent memory falls into four categories:
Most systems only implement short-term and long-term. Episodic and semantic memory require more infrastructure but provide significantly better performance over time.
February 21, 2026
Model Context Protocol (MCP) is Anthropic’s open standard for connecting AI agents to external tools and data. Instead of every agent framework inventing its own tool integration format, MCP provides a single protocol that any agent can speak.
An agent that supports MCP can discover tools at runtime, understand their inputs and outputs, and invoke them – without hardcoded integration code for each tool.
An MCP server exposes three types of capabilities:
February 21, 2026
When an agent has access to dozens of tools, it needs more than names and descriptions to use them well. It needs to know what inputs each tool expects, what outputs it produces, what other tools or infrastructure must be present, and how expensive or risky a call is. A structured skill definition captures all of this in a machine-readable format.
Function signatures tell you the types of parameters. They do not tell you that a skill requires kubectl to be installed, takes 10-30 seconds to run, needs cluster-admin permissions, and might delete resources if called with the wrong flags. Agents making autonomous decisions need this information up front, not buried in documentation they may not read.