v1.2.0 — Major Architecture Upgrade + Marketplace Support
AI Team OS v1.2.0
🎉 Marketplace Install Now Available!
claude plugin marketplace add CronusL-1141/AI-company
claude plugin install ai-team-os
# Restart Claude Code once — done!⚠️ Existing Users: Please Reinstall
If you installed a previous version (v0.8.0 or earlier), please uninstall and reinstall:
# Uninstall old version
pip uninstall ai-team-os -y
claude plugin uninstall ai-team-os
# Fresh install (choose one):
# Option A: Marketplace (recommended)
claude plugin marketplace add CronusL-1141/AI-company
claude plugin install ai-team-os
# Option B: pip
pip install ai-team-os
python -c "from aiteam.hooks.install import main; main()"
# Option C: AI-Assisted
# Tell Claude Code: "Read https://github.com/CronusL-1141/AI-company/blob/master/INSTALL.md and follow the instructions"What's New (v0.9 → v1.2)
v1.2.0 — Agent Autonomy & Reliability
- 🤖 Watchdog Heartbeat — Agent health monitoring with 5-min TTL
- 📊 SRE Error Budget — 4-level response model (GREEN/YELLOW/ORANGE/RED)
- ✅ Completion Verifier — Prevents "hallucinated completion"
- 🔌 Auto Port Discovery — No more port conflicts between projects
- 🌐 MCP HTTP Endpoint —
/mcp/bonus endpoint for external clients - 📦 Ecosystem Recipes — GitHub, Slack, Linear integration guides
- 🛡️ Startup Lock TTL — Auto-cleanup stale locks after 60s
v1.1.0 — Trust & Intelligence
- ⭐ Agent Trust Scoring (0-1 scale, auto-adjusted)
- 💾 Semantic Cache (BM25+Jaccard, 40-73% cost reduction)
- 📋 Tool Tiering (15 core + 46 advanced)
v1.0.0 — CC-First Reliability
- 🔒 File Lock / Workspace Isolation
- 🔄 Error Recovery Mapping (auto-recovery hints on all 107 tools)
- 💬 Channel Messaging with @mention
- 🔧 Git Automation (auto-commit, create-pr, status-check)
- 🛡️ Guardrails L1 (input validation, PII detection)
- 📝 Execution Pattern Memory (learn from success/failure)
- 🗄️ Alembic Database Migration
v0.9.0 — Foundation
- 🔍 BM25 Search (3-5x better than keyword matching)
- 📊 Prompt Registry (version tracking + effectiveness stats)
- 📝 Event Log Enhancement (entity tracking + state snapshots)
- 🏗️ Server Modular Split (3050 → 57 lines entry + 14 modules)
- ⚡ Session Startup Optimization (15-25s → 1-2s)
- 🗣️ Debate Mode (4-round structured: Advocate→Critic→Response→Judge)
- 📊 3 Dashboard Pages (Pipeline, Failures, Prompts)
Stats
- 107 MCP Tools | 25 Agent Templates | 9 Hook Events
- 18 Dashboard Pages | 769+ Tests
- Python 3.11+ | Claude Code Compatible
Known Limitations
- Install to system Python, NOT inside a project .venv
- First marketplace install requires one CC restart
- Dashboard available at localhost:8000 after API starts