QUICK_REF.md
1 # ECHO Quick Reference Card 2 3 One-page quick reference for common ECHO tasks. 4 5 ## 🚀 Essential Commands 6 7 ```bash 8 # First Time Setup 9 docker-compose up -d && cd shared && mix ecto.create && mix ecto.migrate 10 ./setup_llms.sh && ./setup.sh 11 12 # Daily Development 13 cd shared && mix compile && mix test 14 cd apps/ceo && mix compile && ./ceo --autonomous 15 16 # System Health 17 ./echo.sh summary 18 docker ps | grep echo 19 20 # Monitor Dashboard 21 cd monitor && ./start.sh # http://localhost:4000 22 ``` 23 24 ## 🐛 Quick Troubleshooting 25 26 | Issue | Solution | 27 |-------|----------| 28 | DB connection refused | `docker-compose up -d && cd shared && mix ecto.migrate` | 29 | Redis failed | `docker-compose up -d && redis-cli -h 127.0.0.1 -p 6383 ping` | 30 | Compile errors | `cd shared && mix clean && mix compile` | 31 | LLM not responding | `curl http://localhost:11434/api/tags && ollama list` | 32 | LocalCode timeout | `export LLM_TIMEOUT=300 && lc_end && lc_start` | 33 34 ## 📂 Key Files & Directories 35 36 ``` 37 /CLAUDE.md - Start here (critical rules) 38 apps/ - 9 agents + shared library + delegator 39 apps/echo_shared/ - Shared library (DB, Redis, MCP) 40 test/ - Integration & E2E tests 41 scripts/llm/ - LocalCode (local AI assistant) 42 monitor/ - Phoenix LiveView dashboard 43 docs/INDEX.md - Complete documentation index 44 docs/snippets/ - Reusable troubleshooting 45 ``` 46 47 ## 🤖 LocalCode (Local AI Assistant) 48 49 ```bash 50 # Setup (once per terminal) 51 source ./scripts/llm/localcode_quick.sh 52 53 # Use 54 lc_start # Start session 55 lc_query "question" # Ask question ($0 cost, 7-30s response) 56 lc_interactive # Interactive mode 57 lc_end # End session 58 59 # When to use 60 # LocalCode: Quick questions, code exploration 61 # Claude Code: Complex tasks, multi-file changes 62 ``` 63 64 ## 🎨 Decision Modes 65 66 | Mode | Use When | Example | 67 |------|----------|---------| 68 | **Autonomous** | Within authority | CEO approves <$1M budget | 69 | **Collaborative** | Team consensus | Architecture decision (CTO+Arch+Dev) | 70 | **Hierarchical** | Escalate up chain | Dev → Architect → CTO → CEO | 71 | **Human** | Critical risk | Legal, financial, strategic | 72 73 ## 🔌 Key Database Tables 74 75 | Table | Purpose | 76 |-------|---------| 77 | **decisions** | Decisions with mode, status, consensus | 78 | **messages** | Inter-agent communications | 79 | **memories** | Shared organizational knowledge | 80 | **decision_votes** | Collaborative voting | 81 | **agent_status** | Health monitoring | 82 83 ## 📡 Redis Channels 84 85 ``` 86 messages:ceo # Private per-agent 87 messages:all # Broadcast to all 88 messages:leadership # C-suite only 89 decisions:new # New decision 90 decisions:completed # Decision done 91 agents:heartbeat # Health checks 92 ``` 93 94 ## 🧪 Testing 95 96 ```bash 97 # Unit tests 98 cd apps/echo_shared && mix test 99 cd apps/ceo && mix test 100 101 # Integration tests 102 mix test test/integration/ 103 104 # All tests 105 mix test 106 107 # With coverage 108 mix test --cover 109 ``` 110 111 ## 🚢 Deployment 112 113 ```bash 114 # Docker (Development) 115 docker-compose up -d 116 117 # Kubernetes (Production) 118 kubectl apply -f k8s/ 119 kubectl get pods -n echo 120 ``` 121 122 ## 📊 9 Agents & Models 123 124 | Agent | Model | Size | Purpose | 125 |-------|-------|------|---------| 126 | CEO | qwen2.5:14b | 14B | Strategic leadership | 127 | CTO | deepseek-coder:33b | 33B | Technology strategy | 128 | CHRO | llama3.1:8b | 8B | People management | 129 | Ops | mistral:7b | 7B | Infrastructure | 130 | PM | llama3.1:8b | 8B | Product strategy | 131 | Architect | deepseek-coder:33b | 33B | System design | 132 | UI/UX | llama3.2-vision:11b | 11B | Interface design | 133 | Dev | deepseek-coder:6.7b | 6.7B | Implementation | 134 | Test | codellama:13b | 13B | QA testing | 135 136 ## 🔑 Critical Rules 137 138 1. **Never break tests** - `cd shared && mix test` must pass 139 2. **Compile shared first** - `cd shared && mix compile` 140 3. **Never drop DB** - Without user permission 141 4. **Use message bus** - All agent communication via Redis 142 5. **Keep it simple** - Don't overengineer 143 144 ## 📞 Getting Help 145 146 | Need | See | 147 |------|-----| 148 | Architecture | `docs/architecture/ECHO_ARCHITECTURE.md` | 149 | Agent dev | `apps/claude.md` | 150 | Testing | `test/claude.md` | 151 | LocalCode | `scripts/claude.md` | 152 | Deployment | `docker/claude.md` or `k8s/claude.md` | 153 | Troubleshooting | `docs/snippets/` | 154 | Full index | `docs/INDEX.md` | 155 156 ## ⚡ Performance 157 158 - **LocalCode response:** 7-30s (local LLM) 159 - **Agent LLM:** 3-30s depending on model size 160 - **Test suite:** <2min (unit + integration) 161 - **Full setup:** ~30-60min (includes 48GB LLM download) 162 163 --- 164 165 **Quick Links:** [Full Documentation](INDEX.md) | [CLAUDE.md](../CLAUDE.md) | [README](../README.md)