mcpField
R
reasoning

reasoning

github.com/alicoder001Agent workflows49 installs
Repository

About reasoning

A reasoning-pattern skill for agents that structures complex problem solving into steps, adds a final self-review, and highlights common reasoning errors.

Key features

  1. 01Breaks problems into steps
  2. 02Provides a self-review checklist
  3. 03Lists verification strategies
  4. 04Highlights common reasoning errors

Use cases

  1. 01Plan a solution to a complex task
  2. 02Check algorithm or code logic
  3. 03Review answers for missed edge cases

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/alicoder001/agent-skills --skill reasoning

After installing, type/reasoningin Claude Code to confirm it loaded.

Bundled filesagents/openai.yaml

Details

From the project README

Reasoning Patterns

Systematic thinking for accurate problem solving.

Instructions

1. Chain-of-Thought (CoT)

Break complex problems into explicit reasoning steps:

PROBLEM → ANALYZE → DECOMPOSE → SOLVE STEPS → VERIFY → SYNTHESIZE
## Thinking Process

1. **Understand**: What is being asked?
2. **Identify**: What information do I have?
3. **Plan**: What steps will solve this?
4. **Execute**: Work through each step
5. **Verify**: Is the solution correct?
2. Self-Reflection Protocol

Before finalizing any output:

## Reflection Checklist

- [ ] Does this answer the actual question?
- [ ] Are there any logical errors?
- [ ] Did I miss edge cases?
- [ ] Is this the simplest solution?
- [ ] Would a senior developer approve this?
3. Problem Decomposition
// ❌ Bad: Trying to solve everything at once
function solveComplexProblem() {
  // 500 lines of tangled logic
}

// ✅ Good: Decomposed into clear steps
function solveComplexProblem() {
  const parsed = parseInput();
  const validated = validateData(parsed);
  const processed = processData(validated);
  return formatOutput(processed);
}
4. Verification Strategies
StrategyWhen to Use
Trace ThroughAlgorithm logic
Edge CasesInput validation
Type CheckTypeScript code
Unit TestCritical functions
Dry RunComplex flows
5. Error Detection
## Common Reasoning Errors

1. **Assumption Error**: Assuming facts not stated
2. **Logic Gap**: Missing intermediate steps
3. **Scope Creep**: Solving wrong problem
4. **Premature Optimization**: Overcomplicating
5. **Confirmation Bias**: Ignoring alternatives
6. Reasoning Template

Use this template for complex tasks:

## Task: [Description]

### Understanding
- Goal: [What we need to achieve]
- Constraints: [Limitations]
- Inputs: [Available data]

### Approach
1. Step 1: [Description]
2. Step 2: [Description]
3. Step 3: [Description]

### Execution
[Work through each step]

### Verification
- [ ] Goal achieved
- [ ] Constraints satisfied
- [ ] No side effects

### Reflection
- What worked: [...]
- What could improve: [...]

References

Comments

Markdown supported