Skip to main content

​
January 2026

​
v0.5.1 - Polish & Performance

Production hardening with cross-platform fixes and performance improvements.
Features:
  • squads status - Sprint goals display in status view
  • squads trigger - Time-windowed execution statistics
Improvements:
  • 385% faster squads help via lazy-loading Redis
  • Cross-platform path handling for cron/autonomous commands
  • Clearer budget display for different subscription plans
  • Helpful messages when required arguments are missing
Security:
  • Removed hardcoded database credentials
  • Added DEBUG logging to error handlers

​
v0.5.0 - Production Deployment

Ready for production with Docker, distributed locking, and guided onboarding.
Features:
  • squads setup - Guided interactive onboarding with template system
  • squads cron - Native launchd scheduling for macOS
  • squads workers --all - Kill all agent tmux sessions
  • Zombie detection and cleanup for orphaned sessions
Infrastructure:
  • Docker production deployment configuration
  • Redis distributed locking for safe concurrent access
  • Local cooldown checks for scheduled runs
  • BRIDGE_API injection for human-in-the-loop approvals

​
v0.4.13 - Multi-LLM Support

Use different LLM providers by delegating to their native CLIs.
Features:
  • squads providers - Show available LLM CLIs and installation status
  • squads autonomous - Autonomous scheduling command
  • squads slack sync - Slack channel synchronization
  • squads dash - Declarative dashboard engine
  • --provider flag for squads run - Override provider per execution
  • SQUAD.md providers: block - Configure squad-level defaults
  • Agent-level provider: - Override in agent frontmatter or header
  • Support for 7 providers: Anthropic, Google, OpenAI, Mistral, xAI, Aider, Ollama
Provider Resolution:
  1. Agent file (provider: xai)
  2. CLI flag (--provider=google)
  3. Squad default (providers.default)
  4. Fallback to anthropic
Improvements:
  • squads init --yes - Non-interactive mode with auto-detection
  • Auto-install Claude CLI with web-based login
  • Memory --auto-learn flag for automatic learning capture
  • Telemetry with userType segmentation
Philosophy: Unix-style composition. Each provider maintains their own CLI. We orchestrate.

​
v0.4.11 - Learning Loop

Autonomous learning that compounds across sessions.
Features:
  • squads learn - Capture insights that persist across sessions
  • squads learnings show - View squad learnings with filtering
  • squads learnings search - Search across all squads
  • Auto-tagging and category inference
  • Stop hook prompts for learning capture
  • /squads-learn skill for Claude auto-invocation
Categories: success, failure, pattern, tip

​
v0.4.10 - Execution Contexts

Features:
  • Dynamic MCP configuration per squad
  • Model and budget settings in SQUAD.md
  • SQUAD.md parser exported for programmatic use

​
December 2025

​
v0.1.0 - Initial Release

First public release of Agents Squads.
Features:
  • squads CLI for managing agents and squads
  • Markdown-based agent definitions
  • Persistent memory system
  • Goal tracking and feedback collection
  • Integration with Claude Code
Commands:
  • squads init - Initialize new projects
  • squads run - Execute agents and squads
  • squads status - View squad status
  • squads memory - Query and manage memory
  • squads goal - Track squad goals
  • squads feedback - Record execution feedback

​
Roadmap

​
Q1 2026

  • Multi-model support (7 providers) - v0.4.13
  • CLI dashboard (squads dash) - v0.4.13
  • Production deployment (Docker) - v0.5.0
  • REST API for programmatic access
  • Node.js and Python SDKs
  • Webhook support for event notifications

​
Q2 2026

  • Visual web dashboard
  • Team collaboration features
  • Enterprise SSO
  • Enhanced memory search

​
Versioning

We follow Semantic Versioning:
  • Major: Breaking changes
  • Minor: New features (backwards compatible)
  • Patch: Bug fixes

​
Stay Updated

GitHub Releases

Watch for new releases

Newsletter

Subscribe for updates