While AI coding agents like Claude Code and Codex excel at breaking down tasks into structured workflows, they frequently fail at creative ideation. Trained to favor high-frequency patterns, agents consistently default to safe, repetitive answers when tasked with brainstorming. Furthermore, running multiple possibilities inside a single context window causes ideas to bleed together, degrading the quality of thought. The solution to this limitation is the “ADHD skill”—a tool that scatters the agent’s reasoning across isolated context windows using a Tree of Thought framework.

How the ADHD Skill Enhances AI Ideation

The ADHD skill transforms how AI agents generate and evaluate ideas by breaking away from linear thinking. Key architectural components include:

• Tree of Thought Architecture: Instead of processing options sequentially in one context, the skill spins up multiple sub-agents in parallel. Each agent works in complete isolation inside its own context window without knowledge of the others.
• Frame Library: Each sub-agent views the problem through a unique “frame” or prompt lens, forcing distinctly different perspectives on the same issue.
• Strict Critic Agent: An isolated critic agent—prompted as a skeptical senior engineer—scores generated ideas based on three criteria: Novelty (creativity/uniqueness), Viability (realistic execution), and Fit (alignment with the core problem).
• Token Optimization Pre-check: To prevent excessive token consumption, auto-invoked requests undergo a pre-check evaluation. The skill only executes if the problem is open-ended, carries high stakes, and isn’t phrased as a standard or quick task.

Key Practical Applications

The skill does not implement final code or content directly; rather, it formulates deep strategic options that developers can then direct the primary agent to execute. Two key areas benefit significantly from this approach:

• Test-Driven Development (TDD): Standard agents usually write basic tests covering obvious logic paths. Using the ADHD skill across TDD prompts forces the agent to explore multi-branch strategies, uncovering subtle edge cases and performance bottlenecks before application code is written.
• Pre-Launch UX & Churn Evaluation: Running the skill on an almost-ready application allows agents to analyze the user experience from multiple angles. It flags potential churn points, usability traps, and discrepancies between product requirement documents (PRDs) and actual implementation before public launch.

Key Takeaways

By preventing context contamination and actively forcing AI past its standard top-three responses, the ADHD skill transforms safe, generic agent outputs into non-obvious, deeply analyzed strategies. It is particularly effective for high-stakes technical architecture, complex test suites, and product launches where subtle oversights can lead to major failures.

Mentoring question

How could integrating a multi-frame ‘Tree of Thought’ evaluation alter the way your team uses AI agents for architecture planning and edge-case testing?

Source: https://youtube.com/watch?v=m6IXL_YGqBQ&is=3kyH_2Lv7Osutg99