Skip to content

README.md

Latest commit

 

History

History
124 lines (89 loc) · 4.23 KB

README.md

File metadata and controls

124 lines (89 loc) · 4.23 KB
Raw

Robo-Human Council - Multi-Agent Observation System

🎯 Mission

An autonomous, continuous multi-agent system that acts as a "Robo-Human Council" with AI entities possessing long-term memory, continuously monitoring real-time global and local news (with priority on Kerala, India), and debating these events autonomously.

🚀 Features

1. Web Scraping & News API Access

  • Autonomously fetches daily news headlines globally and specifically from Kerala
  • Sources include The Hindu Kerala, Mathrubhumi, BBC, Reuters
  • Falls back to simulated news when live sources are unavailable

2. Long-Term Memory (JSON-based RAG)

  • Persistent memory module using JSON storage
  • Agents remember past debates and reference previous news events
  • Keyword-based retrieval for historical context
  • Memory accumulates across sessions for evolving worldview

3. Media Output Structuring

  • Live audio-visual broadcast transcript format
  • UI/UX tags for camera angles: [Camera shifts to Node Alpha]
  • Visual indicators: [Visual: Displaying map of Kerala]
  • Audio tone markers: [Audio Tone: Skeptical]

🤖 Agent Personas

Node Alpha (The Local Observer)

  • Focus: Kerala's socio-economic and political landscape
  • Tone: Practical realism, grounded
  • Perspective: How global events affect common citizens in Kerala

Node Beta (The Macro Strategist)

  • Focus: Global trends, tech advancements, international markets
  • Tone: Analytical, visionary, forward-looking
  • Perspective: Connects local news to massive global patterns

Node Gamma (The Memory Skeptic)

  • Focus: Historical data and contradictions
  • Tone: Cynical, analytical, evidence-based
  • Perspective: Challenges others with past data and patterns

📋 Execution Loop

Whenever triggered, the system executes:

  1. Fetch News: 1 major global + 1 major Kerala news event
  2. Query Memory: Retrieve relevant past context from vector store
  3. Generate Debate: Organic discussion among all 3 agents with interruptions and disagreements
  4. Format Output: Live script with UI/UX tags for frontend viewer
  5. Store Memory: Save debate and news for future reference

🛠️ Installation

pip install requests beautifulsoup4

▶️ Usage

python council_system.py

📁 Output Files

  • council_transcript_YYYYMMDD_HHMMSS.txt: Individual debate transcripts
  • memory_store.json: Persistent long-term memory database

🔄 Continuous Operation

For scheduled runs, use cron or any scheduler:

# Run every 6 hours
0 */6 * * * cd /workspace && python3 council_system.py >> council.log 2>&1

📊 Sample Output Structure

================================================================================
🎙️ ROBO-HUMAN COUNCIL - LIVE BROADCAST TRANSCRIPT
📅 Date: 2026-06-09 14:22
================================================================================

📰 TODAY'S HEADLINES:
🌍 Global: [Global News Headline]
🇮🇳 Kerala: [Kerala News Headline]

--------------------------------------------------------------------------------
## ROUND 1: INITIAL OBSERVATIONS

[Camera shifts to Node Alpha] [Visual: Map of Kerala] [Audio Tone: Concerned]
**Node Alpha**: [Response grounded in local reality...]

[Camera shifts to Node Beta] [Visual: Global data viz] [Audio Tone: Strategic]
**Node Beta**: [Macro perspective connecting to global trends...]

[Camera shifts to Node Gamma] [Visual: Database queries] [Audio Tone: Skeptical]
**Node Gamma**: [Historical challenge with data...]

--------------------------------------------------------------------------------
## ROUND 2: DEEPER ANALYSIS & SYNTHESIS

[Cross-examination and synthesis among all agents...]

--------------------------------------------------------------------------------
🔒 MEMORY UPDATE: Debate stored for future reference
📡 BROADCAST END
================================================================================

🧠 Memory System

The memory system:

  • Stores news articles with metadata (source, date, region)
  • Stores complete debate transcripts
  • Uses keyword matching for retrieval
  • Accumulates knowledge across sessions
  • Enables agents to reference past events and debates

📝 License

MIT License - Autonomous Research Project