big-three

What problem does it solve?

This Skill prevents low-quality agent outputs by forcing clear decisions about the model, the instructions (prompt), and the information available (context).

Core Features & Use Cases

• Model selection (Model layer): Maps planning/implementation/execution to the appropriate Claude tier (e.g., Opus for planning, Sonnet for building, Haiku for fast tasks).
• Prompt specificity (Prompt layer): Recommends structured, verifiable prompts that include validation steps instead of vague directives.
• Context minimization (Context layer): Enforces minimum viable context and boundary compression so the agent sees only what matters, with recency prioritized.

Use Case: You need an autonomous repo-analysis workflow; apply the Big Three to ensure the orchestrator plans correctly, the specialist implements with clear verification, and the worker executes discrete steps using only necessary context.

Quick Start

Ask the agent to produce a task plan using the Model-Prompt-Context framework, then self-check in the order Model, Prompt, Context before executing.