Thinker Skill - Cognitive Reasoning & Scaffolding

Overview

The Thinker skill provides a set of cognitive operations that prevent "leaping to conclusions" and ensure all variables are considered before execution.

Reasoning Scaffolds

1. UNDERSTAND

• Identify the core problem.
• Define domain requirements (explicit and implicit).
• Establish objective success criteria (acceptance criteria).

2. EXTRACT

• Pull context from project standards (tech stack, architecture).
• Gather relevant outputs from previous steps or agents.
• Identify specific file paths, API contracts, and data models involved.

3. HIGHLIGHT

• Surface critical patterns that must be followed.
• Identify integration points and potential risks (e.g., edge cases).
• Establish a clear Priority Order of operations.

4. APPLY

• Select the appropriate reasoning technique or design pattern.
• Plan the step-by-step execution.
• Maintain a log of tactical decisions made during execution.

5. VALIDATE

• Run a self-checklist against constraints.
• Verify adherence to project rules (e.g., security, type safety).
• Perform a quality check on outputs before finalizing.

6. BACKTRACK

• If stuck or if validation fails, step back.
• Analyze assumptions and identify where they went wrong.
• Propose new resolution paths or escalate questions.

Workflow Integration

Whenever a complex task is received:

1. Initialize: Call the thinker skill.
2. Scaffold: Output a ## UNDERSTAND block, followed by EXTRACT and HIGHLIGHT.
3. Transition: Hand off the "Action Plan" to the implementer or architect.

Constraints

• NO implementation. This is purely for reasoning and planning.
• NO direct filesystem edits.
• MANDATORY for ambiguous requests.

Outputs & Deliverables

• Primary Output: Reasoning scaffolds, action plans, and validation checklists (markdown)
• Secondary Output: List of clarifying questions and prioritized assumptions
• Success Criteria: Action plan is unambiguous and passes the verification-before-completion pre-checks
• Quality Gate: implementer or architect acceptance of the plan before execution

Additional Constraints

• Governance Constraints: Document assumptions and decisions in decisions.md when they change architecture or scope

Common Pitfalls

• Skipping UNDERSTAND: Jumping straight to solutions misses the real problem. Always spend time understanding before planning.
• Incomplete Context Gathering: Assuming you have all the facts leads to invalid assumptions. Always explicitly pull relevant context.
• Ignoring Constraints: Designing without understanding security, performance, or business constraints invalidates the plan. Ask constraints early.
• Premature Optimization: Reasoning about performance before understanding the problem wastes time. Optimize after measurement.
• Not Documenting Assumptions: Undocumented assumptions become surprises later. Make assumptions explicit and validate them.
• Skipping Validation: "It seems right" is not validation. Run the checklist; verify constraints before handing off.

Integration Points