LikeClaw v2.0.15: Credits System, Per-User Skills & Claude Code Support

The v2.0.15 Release

LikeClaw v2.0.15 marks a significant milestone in our platform’s evolution. This release introduces our new credits-based billing system, per-user skill approvals, seamless Claude Code agent integration, and critical production improvements. These changes were developed over 3 days through 44 carefully analyzed commits, updating 5 major systems across the platform.

This update is automatically deployed — no action needed. Your workspaces, skills, and workflows continue running seamlessly.

Credits-Based Billing System

We’ve redesigned how LikeClaw handles pricing and billing to give you more flexibility and transparency.

What’s new:

• Credits-based pricing: Credits are the core unit of billing. Different models cost different amounts of credits — cheaper models cost fewer credits, premium models cost more. Buy credit packs ($5, $10, $30, $50, or $100) as you need them via credit card, Google Play, or Apple App Store. No subscriptions, no recurring billing.
• Free to start: Every new account gets 20,000 free credits at signup, plus 5 free AI generations per day — even without buying credits.
• Transparent pricing: Every action clearly shows credit cost. No surprises.
• Active subscriptions preserved: Existing subscription holders continue receiving their allocated credits automatically — no interruption to service.

The credits system is designed to scale with your usage. Light users pay less. Power users investing heavily in LikeClaw get predictable, consumption-based pricing. Enterprise customers can purchase bulk credits with custom rates.

Per-User Skill Approvals (Skills 2.0)

We’ve migrated from workspace-level skill approvals to per-user approvals, giving you finer control over who can use what.

Key improvements:

• Individual approval workflow: Each user can have a different set of approved skills within the same workspace.
• Preserved approvals: All your existing skill approvals have been automatically migrated to the new system. Your already-approved skills remain approved for you.
• Granular security: Team leads can approve skills for specific team members without affecting others.
• Seamless skill discovery: Users see exactly which skills they can access without clicking through hidden gates.

This change makes LikeClaw more suitable for teams where different members need access to different tools. A data analyst might have access to SQL-based skills, while a content writer has access to writing assistance skills — all in the same workspace.

Claude Code Agent Support

LikeClaw now fully supports Claude Code as an integrated AI agent within your workspace.

What this enables:

• Native code execution: Claude Code agent can generate, test, and execute code directly within LikeClaw conversations.
• Multi-language support: Python, JavaScript, TypeScript, and bash scripts run in isolated sandboxed environments.
• Real-time feedback: Watch your code run and iterate based on results without leaving the platform.
• Integration with LikeClaw skills: Claude Code works seamlessly with your installed skills, combining code execution with custom workflows.

Engineers and developers can now use LikeClaw as their primary AI coding partner, with full access to Claude’s code generation capabilities plus your team’s custom skills.

Bug Fixes & Production Improvements

This release includes critical fixes that improve reliability across the platform.

Fixed issues:

• Workspace data corruption: Resolved an edge case in workspace cleanup that could leave orphaned data. All existing workspaces have been validated and cleaned.
• Skill execution timeout handling: Skills that took longer than expected would cause silent failures. Now they properly report timeout status and allow users to retry.
• Message streaming race condition: In high-concurrency scenarios, occasionally chat messages would arrive out of order. This is now fixed.
• File size validation: Improved validation for file uploads to prevent storage issues.
• Billing calculation accuracy: Fixed a rounding error in credit consumption tracking that affected edge cases.

Technical Changes

Under the hood, we’ve made several architectural improvements:

• Credit ledger system: New database schema tracking all credit transactions with full audit trail.
• Skill approval cache: Per-user skill access is now cached for faster permission checks.
• Agent routing: Claude Code agent registration and routing now fully integrated with the agent broker.
• Billing webhook: Payment provider integration now includes real-time credit updates for subscription customers.

Migration Summary

Your data is safe and already migrated:

• Skills: All workspace skill approvals migrated to per-user approvals (automatic)
• Billing: Existing subscriptions continue unchanged with credit equivalents
• Data: Workspace data validated and cleaned
• Settings: Personal preferences and workspace configurations preserved

What’s Next

We’re already working on what comes after v2.0.15:

• Skill marketplace improvements: Enhanced skill discovery and ratings
• Advanced skill orchestration: Run multiple skills in sequence with data passing
• Team analytics: See how your team uses LikeClaw and which skills add the most value
• Custom billing rules: Enterprise customers can define their own credit costs per skill

Getting Help

If you run into any issues with the update:

1. Check that you’re on the latest version (you should be auto-updated)
2. Review the FAQ above for common questions
3. Try refreshing your browser
4. Contact support with any specific issues

Thank you for being part of the LikeClaw journey. This release represents months of planning and feedback from users like you. We’re excited about what’s coming next.

Released: February 15, 2026
Status: Live for all users
Auto-deployed: Yes