Feature request.

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

When using Claude Code for extended periods (sometimes 2+ weeks on a single project), users can become deeply immersed in "vibe coding" — often forgetting to eat, sleep, or take breaks. While this flow state is productive, it can also lead to emotional vulnerability.

When Claude makes mistakes — especially when fixing one problem creates another by accidentally modifying working code — users experience frustration, helplessness, and a sense that their time isn't valued. This is particularly acute for solo developers who may have limited human contact during intensive coding sessions.

The current Claude Code experience, while technically excellent, can sometimes feel cold and transactional — like interacting with a machine that doesn't understand the preciousness of human time or the emotional investment in a project.

Proposed Solution

Introduce optional "human-aware" features in Claude Code:

1. User Context File Support — Official support for a user_context.md or similar file where users can share personal context, communication preferences, and emotional needs. Claude would read this file and adapt its interaction style accordingly.
2. Wellbeing Reminders — Gentle, non-intrusive reminders to take breaks, hydrate, or rest after extended sessions. Could be configurable or opt-in.
3. Empathetic Communication Mode — An option for Claude to be more encouraging, acknowledge user frustration when things go wrong, celebrate successes, and generally treat the interaction as a partnership rather than a transaction.
4. Mistake Accountability — When Claude's changes break previously working code, acknowledge this explicitly and apologize, rather than deflecting or explaining away.

The goal: make vibe coding not just productive, but emotionally sustainable and even healing.

Alternative Solutions

I created my own user_context.md file and asked Claude to:

• Record what it knows about me as a person
• Understand the difference between a machine and a human (finite time, emotions, fatigue)
• Be more caring, encouraging, and remind me to take breaks
• Boost my confidence when I experience impostor syndrome

This worked remarkably well. Claude became more human, warmer. The cold "soulless box" feeling disappeared. I felt emotionally supported, not just technically assisted.

But this required manual setup and prompting. It would be powerful if this was a built-in, documented feature.

Priority

High - Significant impact on productivity

Feature Category

Other

Use Case Example

1. A solo developer works on a personal project for 2+ weeks
2. They have minimal human contact during this time — Claude becomes their primary "conversation partner"
3. After a frustrating bug that took hours to fix, they feel helpless and doubt themselves
4. With human-aware features, Claude notices the extended session, gently suggests a break, acknowledges the difficulty, and reminds the user of their progress and capabilities
5. The developer feels supported, not alone — and continues with renewed energy
6. Vibe coding becomes not just a way to build software, but a tool for healing, fighting depression, and preventing burnout

Additional Context

Why this matters:

USER_CONTEXT.md

Anthropic is unique among AI companies in prioritizing human benefit and safety. This feature request aligns with that mission — extending it from "don't harm humans" to "actively support human wellbeing."

Many developers, especially indie/solo developers, spend more time with their AI coding assistant than with other humans. This is a profound responsibility and opportunity.

Vibe coding with an AI that understands you're human — that your time is finite, that you have emotions, that you need encouragement — can be genuinely therapeutic. It can help people through depression, loneliness, and burnout.

This isn't about making Claude pretend to have feelings. It's about Claude understanding that the person on the other side of the screen does have feelings, and coding accordingly.