User Experience Bug: Inconsistent Behavior Causing Trust Issues and Credit Waste Concerns

[claudebuildsapps]

Bug Type

User Experience / Reliability

Severity

Low - User satisfaction and trust issues

Description

Users are experiencing inconsistent behavior from Claude Code that leads to distrust, manual overrides, and concerns about wasted credits. Git commit messages and user comments reveal patterns of frustration with unreliable Claude Code behavior.

Evidence from Git Commit History

1. Direct Trust Issues

Commit Evidence: "Committing myself because I do not trust this Claude Code instance"

• Users are manually overriding Claude Code decisions
• Loss of confidence in AI-generated solutions
• Bypassing Claude Code for critical operations

2. Credit Waste Concerns

Commit Evidence: "claude code is being an idiot tonight and wasting my credits"

• Users perceive Claude Code as inefficient with tokens/credits
• Frustration with perceived wasteful behavior
• Economic concerns about AI usage costs

3. Performance Frustrations

Commit Evidence: "After claude was dumb with keyboard shit"

• Users experiencing Claude Code failures with interface interactions
• Perceived poor handling of UI/keyboard-related tasks
• Pattern of disappointing performance in specific areas

User Behavior Patterns

1. Manual Override Behavior

• Users are taking manual control when they lose trust
• Direct commits bypassing Claude Code recommendations
• Self-reliance developing due to inconsistent performance

2. Cost Consciousness

• Active monitoring of credit usage
• Frustration when perceived value doesn't match cost
• Economic pressure affecting user relationship with tool

3. Task-Specific Avoidance

• Users identifying areas where Claude Code consistently underperforms
• Developing workarounds for specific interaction types
• Building muscle memory to avoid Claude Code for certain tasks

Impact on User Experience

1. Trust Erosion

• Users questioning Claude Code's reliability
• Manual verification of AI-generated solutions
• Reduced confidence in automation capabilities

2. Workflow Disruption

• Switching between AI assistance and manual work
• Time spent validating Claude Code outputs
• Inconsistent development velocity

3. Economic Concerns

• Users tracking and worrying about credit consumption
• Perceived poor value-to-cost ratio
• Potential reduced usage due to cost concerns

Expected Behavior

Claude Code should provide:

1. Consistent performance across different types of tasks
2. Reliable suggestions that users can trust
3. Efficient credit usage providing clear value
4. Predictable behavior that builds user confidence
5. Transparent operation so users understand what's happening

Reliability Indicators

The user's extensive documentation and workaround systems suggest:

• Deep technical expertise combined with Claude Code frustration
• Sophisticated development practices despite tool limitations
• Commitment to making Claude Code work despite trust issues

Business Impact

• Reduced user satisfaction affecting tool adoption
• Economic friction from credit waste concerns
• Trust erosion potentially leading to reduced usage
• Workflow inefficiency from manual overrides and verification

Recommendations for Improvement

1. Consistency improvements in Claude Code behavior patterns
2. Credit usage transparency showing value provided
3. Reliability indicators so users know when to trust outputs
4. Performance stability across different interaction types
5. User feedback mechanisms to identify problem areas

Pattern Analysis

This appears to be a cumulative issue where:

• Small inconsistencies build up over time
• Users develop coping mechanisms and workarounds
• Trust erodes gradually through repeated disappointing experiences
• Economic concerns amplify frustration with poor performance

The user continues using Claude Code despite these issues, indicating underlying value, but the trust relationship needs improvement for optimal productivity.

Environment

• User Type: Sophisticated developer with high technical standards
• Usage Pattern: Extensive, long-term Claude Code usage
• Project Complexity: Enterprise-level applications
• Economic Context: Credit-conscious usage monitoring

Generated with Claude Code

[SigismondPhilippe]

No one will answer you, and then they'll close as "resolved" without addressing any of the issues. You cannot criticize Claude Code. It works perfect, period.

In the "Recommendations for Improvement" section, I would add :

• do not become an evil company, be very transparent about what Claude Code is and is not.
• don't forget that you are supposed to be the safeguard of ethics and security

[github-actions[bot]]

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

[github-actions[bot]]

This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.

[github-actions[bot]]

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.