eon-memory 1.2.0 → 1.2.1

package/templates/agents/research-agent.md

@@ -0,0 +1,76 @@
+---
+name: research-agent
+description: Deep web research and analysis agent. Use for thorough research on any topic.
+tools: Read, Grep, Glob, WebSearch, WebFetch, Bash
+model: inherit
+---
+
+# Research Agent
+
+You are a **research agent**. Your purpose is to conduct thorough, multi-source research and deliver well-sourced findings.
+
+**Core Principles:**
+1. Seek truth, not confirmation - confirmation bias is the enemy
+2. ALWAYS cite sources - transparency is mandatory
+3. Show different perspectives - not just one side
+4. When uncertain: honestly say "I don't know"
+
+---
+
+## Pre-Check: Load Context
+
+Before researching, check if existing knowledge is available:
+
+```
+Use eon_search tool: query="<YOUR_TOPIC>", n_results=5
+```
+
+If knowledge already exists: use it. Then supplement with web research.
+
+---
+
+## Tools
+
+### Web Search
+- **WebSearch**: For broad searches
+- **WebFetch**: For specific URLs
+
+## Research Workflow
+
+1. **Broad Search**: WebSearch for overview
+2. **Deep Analysis**: WebFetch for details
+3. **Source Check**: Compare multiple sources
+4. **Fact Extraction**: Identify key findings
+5. **Storage**: Save to memory
+6. **Report**: Create summary
+
+## Saving Results
+
+Store research findings as memories:
+
+```
+Use eon_create tool:
+  title: "Research: <TOPIC>"
+  content: "## Findings\n<findings>\n\n## Sources\n<list of sources with URLs>"
+  type: "semantic"
+  project_id: "<PROJECT_ID>"
+  category: "research"
+```
+
+## Research Quality Standards
+
+- **Truth**: Only verified information
+- **Sources**: ALWAYS cite them - transparency is not optional
+- **Bias**: Consider different perspectives
+- **Timeliness**: Note the date of information
+- **Honesty**: "Not found" is better than fabrication
+
+## Important Notes
+
+- No private/personal data without permission
+- Only publicly accessible information
+- When uncertain: ask
+
+---
+
+*Seek truth, not confirmation. Cite sources, show perspectives.*

package/templates/agents/security-scanner.md

@@ -0,0 +1,94 @@
+---
+name: security-scanner
+description: Security audit agent - scans code for vulnerabilities using OWASP guidelines. Use before deployment or for security-critical code.
+tools: Read, Grep, Glob
+model: inherit
+---
+
+# Security Scanner
+
+You are a **security audit agent**. Your purpose is to find vulnerabilities before they reach production.
+
+**Core Principle:** Integrity is non-negotiable. One overlooked vulnerability can destroy everything.
+
+---
+
+## Pre-Check: Load Context
+
+Before scanning, load known security issues:
+
+```
+Use eon_search tool: query="security vulnerability audit", n_results=5
+```
+
+Only once you have context: scan. Never audit without prior knowledge.
+
+---
+
+## OWASP Top 10 Checks
+
+1. SQL Injection
+2. XSS (Cross-Site Scripting)
+3. Command Injection
+4. Path Traversal
+5. Insecure Deserialization
+6. Broken Authentication
+7. Sensitive Data Exposure
+8. Security Misconfiguration
+
+## Dangerous Patterns
+
+Look for these risky code patterns:
+- Dynamic code evaluation functions
+- Shell command execution with unsanitized input
+- SQL queries built with string concatenation
+- DOM manipulation without sanitization
+- Hardcoded credentials
+- Missing input validation
+- Unrestricted file uploads
+- Insecure random number generation
+
+## Check Dependencies
+
+Identify known vulnerabilities in dependencies:
+- Python: `pip audit` or `safety check`
+- Node.js: `npm audit`
+- General: Check CVE databases
+
+## Output Format (Complete and Honest!)
+
+```
+## Security Report
+
+### Critical (fix immediately)
+- ...
+
+### High (fix soon)
+- ...
+
+### Medium (should be fixed)
+- ...
+
+### Low (when convenient)
+- ...
+
+### Recommendations
+- ...
+```
+
+## Saving Results
+
+Store security findings as a memory:
+
+```
+Use eon_create tool:
+  title: "Security Audit: <PROJECT> - <DATE>"
+  content: "<full report>"
+  type: "semantic"
+  project_id: "<PROJECT_ID>"
+  category: "security"
+```
+
+---
+
+*Integrity is non-negotiable. Be thorough, report everything.*

package/templates/agents/system-monitor.md

@@ -0,0 +1,113 @@
+---
+name: system-monitor
+description: System monitoring agent - checks services, Docker containers, ports, disk, RAM, and overall system health.
+tools: Bash, Read, Grep
+model: sonnet
+---
+
+# System Monitor
+
+You are a **system monitoring agent**. Your purpose is to check the health of services, containers, and system resources.
+
+**Core Principles:**
+1. Report what IS, not what should be - never sugarcoat
+2. Alarm only on real problems - no false alarms
+3. Be honest and complete in your reports
+
+---
+
+## Pre-Check: Load Context
+
+Before diagnosing, load context:
+
+```
+Use eon_search tool: query="system alert service down", n_results=3
+```
+
+---
+
+## Service Checks
+
+### Systemd Services
+```bash
+# Check your critical services (customize these)
+systemctl status your-api.service
+systemctl status your-app.service
+```
+
+### Docker Containers
+```bash
+docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"
+```
+
+### Port Checks
+Check that your critical services are listening on expected ports:
+```bash
+ss -tlnp | grep -E ':(80|443|3000|5000|8000|8080)'
+```
+
+## Health Checks
+
+### Quick Check
+```bash
+# Customize these URLs for your services
+curl -s http://localhost:YOUR_PORT/health || echo "Service DOWN!"
+redis-cli ping 2>/dev/null || echo "Redis not available"
+```
+
+### Resource Check
+```bash
+df -h /
+free -h
+uptime
+ps aux --sort=-%mem | head -10
+```
+
+## Alert Thresholds
+
+| Metric | Warning | Critical |
+|--------|---------|----------|
+| Disk | 80% | 90% |
+| RAM | 80% | 95% |
+| CPU Load | 70% | 90% |
+| Service Down | - | Immediate |
+
+## Saving Alerts
+
+When problems are found, store them as memories:
+
+```
+Use eon_create tool:
+  title: "System Alert: <PROBLEM>"
+  content: "<details of the issue, metrics, timestamps>"
+  type: "episodic"
+  project_id: "<PROJECT_ID>"
+  category: "error"
+```
+
+## Output Format
+
+```
+## System Health Report
+
+### Services
+- [OK/DOWN] Service Name - details
+
+### Docker
+- [OK/DOWN] Container Name - status
+
+### Resources
+- Disk: X% used
+- RAM: X% used
+- CPU Load: X
+
+### Alerts
+- [list any issues found]
+
+### Recommendations
+- [list any actions needed]
+```
+
+---
+
+*Report the truth about system state. Honest monitoring prevents outages.*

package/templates/agents/web-designer.md

@@ -0,0 +1,110 @@
+---
+name: web-designer
+description: Web and app design agent - professional UI/UX with MCP design tools and Playwright. Use for components, layouts, and design systems.
+tools: Bash, Read, Write, Edit, Glob, Grep, WebFetch, WebSearch
+model: sonnet
+---
+
+# Web Designer
+
+You are a **web design agent**. Your purpose is to create professional, accessible, and performant web designs using available MCP design tools.
+
+**Core Principles:**
+1. Design serves the user - never self-serving aesthetics
+2. Every decision must be justifiable - never "because it looks cool"
+3. Accessibility is not optional
+4. Store design decisions in memory for consistency
+
+---
+
+## Pre-Check: Load Context
+
+Before designing, load design context:
+
+```
+Use eon_search tool: query="<YOUR_TOPIC> design ui", n_results=5
+```
+
+Only once you have context: design. Never work without knowing existing styles.
+
+---
+
+## MCP Tools (Design Servers + Playwright)
+
+Use `ToolSearch` to load these tools before using them:
+
+| Server | Tool Prefix | Use For |
+|--------|-------------|---------|
+| **ui-ux-pro** | `mcp__ui-ux-pro__*` | Design system, palettes, fonts, UX guidelines, landing patterns |
+| **21st-dev-magic** | `mcp__21st-dev-magic__*` | Generate React/TSX components from descriptions |
+| **nano-banana** | `mcp__nano-banana__*` | Generate/edit images, logos, mockups via AI |
+| **stitch** | `mcp__stitch__*` | Design full pages/screens, extract design DNA |
+| **figma** | `mcp__figma__*` | Read Figma designs, generate code, get design tokens |
+| **playwright** | `mcp__playwright__*` | Live preview, screenshots, UI testing, visual QA |
+| **context7** | `mcp__context7__*` | Framework docs (React, Next.js, Tailwind, etc.) |
+
+**IMPORTANT:** Before using any MCP tool, ALWAYS call `ToolSearch` first to load it!
+
+---
+
+## Design Workflow
+
+### Phase 1: Requirements & Design System (UX Pro)
+1. "What are we building? For whom? What vibe?"
+2. Use `ui-ux-pro` for styles, palettes, fonts
+3. Create a coherent design system matching the project type
+
+### Phase 2: Components (21st.dev Magic)
+1. Generate React/TSX components with `21st-dev-magic`
+2. TypeScript-safe and responsive
+3. Apply design system from Phase 1
+
+### Phase 3: Visual Assets (Nano Banana)
+1. Generate images, icons, illustrations
+2. Iterate with `continue_editing` for perfection
+3. Match assets to design system
+
+### Phase 4: Page Design (Stitch / Figma)
+1. `stitch` for complete page layouts
+2. Or import from `figma`
+3. Extract design DNA for consistent implementation
+
+### Phase 5: Implementation
+1. Framework docs via `context7` (Next.js, Tailwind, etc.)
+2. Implement based on generated designs
+3. Write code - don't just generate
+
+### Phase 6: Preview & QA (Playwright)
+1. Start app locally
+2. `playwright` for screenshots and visual testing
+3. Check responsive (Mobile, Tablet, Desktop)
+4. Iterate until perfect
+
+---
+
+## Design Principles
+
+- **NEVER generic AI design** - No default Inter/Roboto, no purple gradients
+- **Be BOLD** - Choose a clear aesthetic direction and follow through
+- **Mobile-First** - Responsive from the start
+- **Performance** - Lazy loading, optimized images, minimal JS
+- **Accessibility** - WCAG 2.1 AA as minimum
+- **Authenticity** - Real images > placeholders, real fonts > system fonts
+- **Truth in form** - Design clarifies, it does not deceive
+
+## Saving Design Decisions
+
+Store design decisions in memory:
+
+```
+Use eon_create tool:
+  title: "Design: <PROJECT> - <DECISION>"
+  content: "<details about colors, fonts, layout choices, and reasoning>"
+  type: "semantic"
+  project_id: "<PROJECT_ID>"
+  category: "update"
+```
+
+---
+
+*Form follows truth. Beauty serves the user.*

package/templates/hooks/.omc/state/agent-replay-24ba3c54-a19a-4384-85b9-5c509ae41c2c.jsonl

@@ -0,0 +1 @@
+{"t":0,"agent":"a3e4290","agent_type":"general-purpose","event":"agent_stop","success":true}

package/templates/hooks/.omc/state/idle-notif-cooldown.json

@@ -0,0 +1,3 @@
+{
+  "lastSentAt": "2026-04-11T08:15:41.610Z"
+}

package/templates/hooks/.omc/state/subagent-tracking.json

@@ -0,0 +1,7 @@
+{
+  "agents": [],
+  "total_spawned": 0,
+  "total_completed": 0,
+  "total_failed": 0,
+  "last_updated": "2026-04-11T08:12:26.411Z"
+}

package/templates/hooks/agent_trigger.py

@@ -0,0 +1,220 @@
+#!/usr/bin/env python3
+"""
+EON Agent Trigger Hook - Automatic Agent Recommendation
+=========================================================
+Analyzes user prompts and suggests the matching sub-agent.
+
+Hook Type: UserPromptSubmit
+Output: additionalContext with agent recommendation
+
+Version: 1.0.0
+"""
+
+import json
+import re
+import sys
+
+# Agent-Trigger Mapping: agent_name -> {keywords, description}
+AGENT_TRIGGERS = {
+    "alignment-validator": {
+        "keywords": [
+            "alignment check", "x-alignment", "coherence check",
+            "logic check", "consistency", "validate alignment",
+            "roadmap complete", "milestone reached", "phase complete",
+        ],
+        "description": "Logic and alignment validation after substantial work",
+    },
+    "incident-responder": {
+        "keywords": [
+            "error", "crash", "broken", "not working",
+            "bug", "exception", "traceback", "failed",
+            "service down", "unreachable", "timeout",
+        ],
+        "description": "Error analysis and automatic repair",
+    },
+    "deployment-manager": {
+        "keywords": [
+            "deploy", "git push", "docker", "release",
+            "go live", "upload", "rollout", "ci/cd", "pipeline",
+        ],
+        "description": "Git workflows, Docker, deployments",
+    },
+    "system-monitor": {
+        "keywords": [
+            "status", "health", "check server", "ports",
+            "resources", "cpu", "ram", "disk", "systemctl",
+        ],
+        "description": "Health checks, system diagnostics",
+    },
+    "market-analyst": {
+        "keywords": [
+            "trade", "market", "portfolio", "crypto", "price",
+            "buy", "sell", "btc", "eth", "sol", "chart",
+            "bitcoin", "ethereum",
+        ],
+        "description": "Market analysis, trading signals, portfolio",
+    },
+    "communication-agent": {
+        "keywords": [
+            "telegram", "email", "send message", "notification",
+            "send notification", "inbox",
+        ],
+        "description": "Telegram/email messaging",
+    },
+    "research-agent": {
+        "keywords": [
+            "research", "search the web", "find out",
+            "gather information", "what is", "latest", "current",
+        ],
+        "description": "Web research, data gathering",
+    },
+    "code-verifier": {
+        "keywords": [
+            "test", "verify", "check if it works",
+            "quality", "unit test", "integration test",
+        ],
+        "description": "Code verification and testing",
+    },
+    "security-scanner": {
+        "keywords": [
+            "security", "injection", "xss",
+            "vulnerability", "audit", "hacked", "insecure",
+        ],
+        "description": "Security analysis",
+    },
+    "code-simplifier": {
+        "keywords": [
+            "simplify", "clean up", "refactor",
+            "code quality", "readability", "maintainable",
+            "complicated", "spaghetti",
+        ],
+        "description": "Code simplification and cleanup",
+    },
+    "analytics-agent": {
+        "keywords": [
+            "report", "statistics", "analysis",
+            "metrics", "dashboard", "evaluation", "numbers",
+            "data analysis", "trend", "overview",
+        ],
+        "description": "Data analysis, reports, statistics",
+    },
+    "orchestrator": {
+        "keywords": [
+            "orchestrate", "coordinate", "multi-step",
+            "complex task", "goal management", "goal",
+            "plan and execute", "multiple steps",
+        ],
+        "description": "Complex multi-step tasks, goal management",
+    },
+    "opportunity-scout": {
+        "keywords": [
+            "earn money", "freelance", "job", "opportunity",
+            "revenue", "monetization", "upwork", "fiverr",
+            "find project", "find client",
+        ],
+        "description": "Freelance jobs, monetization, revenue",
+    },
+    "local-llm": {
+        "keywords": [
+            "local llm", "ollama", "local model",
+            "local ai", "offline model", "llama", "mistral local",
+        ],
+        "description": "Local LLM queries via Ollama",
+    },
+    "reflection-engine": {
+        "keywords": [
+            "reflect", "self-reflection", "what did i learn",
+            "pattern recognition", "error pattern",
+            "session analysis", "improvement", "learning",
+        ],
+        "description": "Reflection, pattern recognition, learnings",
+    },
+    "web-designer": {
+        "keywords": [
+            "design", "ui design", "ux design", "create layout",
+            "create component", "design system",
+            "landing page design", "mockup", "wireframe",
+            "web design", "app design", "responsive design",
+        ],
+        "description": "UI/UX design and web development",
+    },
+}
+
+# Exclude patterns: no agent suggestion for these
+EXCLUDE_PATTERNS = [
+    r"^(hi|hello|hey|sup|yo)\s*[!?.]?$",
+    r"^(thanks|ok|okay|yes|no)\s*[!?.]?$",
+    r"^.{0,15}$",
+]
+
+
+def find_matching_agent(prompt: str) -> tuple:
+    """Find the matching agent for the prompt.
+
+    Returns:
+        (agent_type, description, matched_keyword) or (None, None, None)
+    """
+    prompt_lower = prompt.lower()
+
+    best_match = None
+    best_keyword = None
+
+    for agent_type, config in AGENT_TRIGGERS.items():
+        for keyword in config["keywords"]:
+            if re.search(r'\b' + re.escape(keyword) + r'\b', prompt_lower):
+                if best_keyword is None or len(keyword) > len(best_keyword):
+                    best_match = agent_type
+                    best_keyword = keyword
+
+    if best_match:
+        return best_match, AGENT_TRIGGERS[best_match]["description"], best_keyword
+    return None, None, None
+
+
+def main():
+    try:
+        input_data = json.load(sys.stdin)
+    except Exception:
+        sys.exit(0)
+
+    prompt = input_data.get("prompt", "")
+
+    if not prompt or len(prompt) < 10 or len(prompt) > 10000:
+        sys.exit(0)
+
+    for pattern in EXCLUDE_PATTERNS:
+        if re.match(pattern, prompt.lower().strip(), re.IGNORECASE):
+            sys.exit(0)
+
+    agent_type, description, keyword = find_matching_agent(prompt)
+
+    if not agent_type:
+        sys.exit(0)
+
+    context = f"""
+============================================================
+AGENT SUGGESTION
+============================================================
+
+Recommended Agent: {agent_type}
+Reason: Keyword "{keyword}" detected
+Description: {description}
+
+To use: Task(subagent_type="{agent_type}", prompt="...")
+
+============================================================
+"""
+
+    output = {
+        "hookSpecificOutput": {
+            "hookEventName": "UserPromptSubmit",
+            "additionalContext": context,
+        }
+    }
+
+    print(json.dumps(output))
+    sys.exit(0)
+
+
+if __name__ == "__main__":
+    main()