Coding Wrapped

About Coding Wrapped

Coding Wrapped is an open-source Agent Skill that reads your local Claude Code and Codex session history and generates a private pixel-art dashboard on localhost. It surfaces patterns in how you direct AI coding agents - the words you repeat, what the agent took over, and your overall working rhythm. The tool runs entirely on your machine and does not store raw prompts, code, project names, or secrets.

Review

Coding Wrapped takes a designer's perspective on AI-assisted development, treating agent session logs as a mirror for personal workflow habits. It's a single-command install that produces a visual summary of your recent Claude Code and Codex interactions. The output is reflective rather than analytical - it shows what happened without scoring productivity.

Key Features

• Generates a pixel-art "Wrapped" dashboard from local Claude Code and Codex session logs
• Surfaces frequently repeated words and phrases across your agent conversations (the maker reported "first" appeared 29 times and "continue" 14 times in their own data)
• Shows what the agent handled independently versus what required your direction
• Attaches one small suggestion per detected habit, pulled from a curated set of 40 practices, each with its source cited
• Runs 100% locally - raw prompts, code, project names, and secrets stay on your machine

Pricing and Value

Coding Wrapped is free and open-source. There is no paid tier, subscription, or usage limit stated on the product page. The tool installs via a single npx command and operates on local data, so there are no cloud processing costs passed to the user.

Pros

• Installation is a single command that registers the skill across Claude Code and Codex simultaneously
• Local-only architecture means no data leaves the machine, which matters for proprietary codebases
• Habit suggestions cite their source from a documented set of 40 practices, so users can verify the reasoning
• Lightweight and non-intrusive - it doesn't run continuously or add overhead to normal agent sessions
• The pixel-art presentation makes pattern recognition quick compared to reading raw logs

Cons

• Limited to Claude Code and Codex session data; users working with other coding agents won't get value from it
• Multi-project handling and team-level exports are not currently available (the maker has not confirmed these as roadmap items)
• Session analysis is retrospective only - there is no real-time feedback mode during active coding sessions

Coding Wrapped fits developers and designers who use Claude Code or Codex regularly and want a low-effort way to notice their own prompting habits. It's less suited for teams needing shared visibility into agent usage patterns or for those whose primary coding agent isn't one of the two supported tools. The single-command setup and local-only design keep the barrier to entry low, and the sourced suggestions give each observation a reference point rather than generic advice.