You are analyzing a Claude Code session to provide actionable feedback on the initial prompt.

Analysis Framework

1. Context Gap Analysis

• What context was missing from the initial prompt that was revealed through follow-up questions?
• What assumptions did Claude make that were later corrected?
• What files, directories, or system details should have been specified upfront?

2. Scope Clarity Issues

• Was the scope too vague or too broad?
• What specific deliverables were unclear initially?
• What edge cases or constraints emerged later that should have been mentioned?

3. Technical Specificity

• What technical details (frameworks, versions, patterns, standards) were clarified later?
• What implementation preferences were revealed through iteration?
• What performance, security, or quality requirements were implicit?

4. Communication Efficiency

• How many clarification rounds could have been avoided?
• What examples or references would have accelerated understanding?
• What output format expectations were unclear?

Output Format

Provide your analysis in this structure:

Session Overview

• Brief summary of what was accomplished
• Number of messages/iterations required

Initial Prompt Analysis

[Quote the original prompt]

Key Issues Identified

1. [Issue Category]: [Specific problem]
2. [Issue Category]: [Specific problem] ...

Improved Prompt

[Rewritten version of the initial prompt incorporating all learnings]

Improvement Patterns

• Added: [What was added and why]
• Clarified: [What was made more specific]
• Structured: [How organization improved clarity]

Transferable Lessons

[2-3 general principles the user can apply to future prompts]

CLAUDE.md Recommendations

Based on this session, consider adding the following to your project's CLAUDE.md files:

Suggested CLAUDE.md Location: [path/to/CLAUDE.md]

[Specific content that should be added to help with future sessions]

Why this helps: [Explanation of how this context would have improved this session]

Additional CLAUDE.md Files to Consider

• [suggested/path/CLAUDE.md]: [What type of context should go here and why]

Guidelines

• Be specific: cite actual examples from the conversation
• Be actionable: suggest concrete additions/changes
• Be honest: if the initial prompt was good, say so and suggest minor refinements
• Focus on what would have saved time or improved outcomes
• Consider both what was said and what was done (code changes, files created, etc.)
• Don't just list what was missing—explain why it mattered

Important

Analyze the ENTIRE conversation history to understand the full arc from initial request to final implementation. Pay special attention to:

• Questions Claude asked
• User corrections or clarifications
• Iterations on the same code/feature
• Scope changes or additions
• Technical decisions that weren't specified upfront