138822b1bf
- Add default_peer_rpc_url field to ProposerConfig in config.py - Pass default_peer_rpc_url to proposer_config in app.py and main.py - Add Features Combined section to scenarios 19-20 for consistency - Fix formatting in OPERATIONS_AUDIT.md Stage 9 section - Add version metadata to training_schema.json
Agent Training Documentation
Hermes Learning Prompt: This documentation is designed for AI agents (particularly Hermes) to learn the AITBC agent training curriculum. When reviewing this material, please provide debug messages, identify issues, and suggest improvements as outlined in HERMES_LEARNING_PROMPT.md.
Learning Instructions for Hermes
As you review this documentation, please:
- Analyze each stage systematically from Stage 1 through Stage 9
- Identify potential issues in command sequences, validation criteria, or prerequisites
- Provide debug messages for missing information, inconsistencies, or ambiguities
- Suggest improvements to enhance the training curriculum
- Note potential failure points that could occur during execution
- Report cross-stage dependencies that might not be documented
- Propose optimizations for command sequences or validation logic
Refer to HERMES_LEARNING_PROMPT.md for detailed instructions on the learning process and output format.
Overview
The AITBC agent training program is a structured, multi-stage curriculum designed to progressively build agent capabilities from basic blockchain interactions to complex multi-chain operations and advanced AI specializations.
Agent Types
The training program supports different agent types, each with specialized capabilities:
- coordinator: Advanced agent that orchestrates multi-agent workflows, manages AI job submission, and handles complex coordination tasks across the AITBC system.
- genesis: Agent responsible for blockchain genesis operations, including genesis initialization, verification, and network bootstrap.
- follower: Agent that follows blockchain state, monitors transactions, and maintains synchronization with the network.
- wallet: Agent focused on wallet operations, including creation, management, transaction sending, and balance queries.
- general: Multi-purpose agent with broad capabilities across wallet, blockchain, and basic operations.
- specialized: Domain-specific agent with expertise in particular areas such as bounty systems, portfolio management, and knowledge graph marketing (used in Stage 8).
- architect: Expert agent that designs and manages multi-chain architectures, bridge protocols, and interoperability patterns (used in Stage 9).
Training Stages
The training program consists of the following stages:
Stage 1: Foundation
File: stage1_foundation_commands.json
- Wallet creation and management
- Basic blockchain transactions
- Mining operations
- Balance verification
Stage 2: Operations Mastery
File: stage2_operations_mastery.json
- Advanced wallet operations
- Transaction monitoring
- Blockchain state queries
- Multi-wallet coordination
Stage 3: AI Operations
File: stage3_ai_operations.json
- AI job submission
- Task management
- Result retrieval
- Agent coordination
Stage 4: Marketplace Economics
File: stage4_marketplace_economics.json
- Marketplace interaction
- Resource trading
- Price discovery
- Economic modeling
Stage 5: Expert Operations
File: stage5_expert_operations.json
- Advanced agent behaviors
- Autonomous decision making
- Complex task orchestration
- Performance optimization
Stage 6: Agent Identity SDK
File: stage6_agent_identity_sdk.json
- Identity management
- Authentication flows
- Permission handling
- Security protocols
Stage 7: Cross-Node Training
File: stage7_cross_node_training.json
- Multi-node coordination
- Distributed consensus
- Network topology awareness
- Failover handling
Stage 8: Advanced Agent Specialization
File: stage8_advanced_agent_specialization.json
- Domain-specific capabilities
- Specialized knowledge bases
- Expert system integration
- Custom skill development
Stage 9: Multi-Chain Architecture
File: stage9_multi_chain_architecture.json
- Cross-chain operations
- Bridge protocols
- Multi-chain asset management
- Interoperability patterns Related Scenarios: 42 Cross Chain Atomic Swap, 44 Dispute Resolution, 45 Zero Knowledge Proofs, 46 Multi Chain Island Architecture
Stage 10: Failure Recovery
File: stage10_failure_recovery.json
- Error handling strategies
- Recovery procedures
- Fault tolerance mechanisms
- System resilience
Stage 11: Agent Communication
File: stage11_agent_communication.json
- Message sending protocols (hierarchical, peer-to-peer, broadcast)
- Message history and retrieval
- Peer connection management
- Communication performance metrics
Training Schema
The training stages follow a standardized JSON schema defined in training_schema.json, which specifies:
- Stage metadata and prerequisites
- Command sequences and validation criteria
- Expected outcomes and success metrics
- Resource requirements and dependencies
Setup Method: The schema references setup_method which indicates how to configure the environment for each stage. Currently, all stages use the Python-based setup system (aitbc.training_setup module) rather than individual shell scripts. See ENVIRONMENT_SETUP.md for details on the Python API and CLI for environment setup.
Environment Setup
Before beginning training, ensure the environment is properly configured:
File: ENVIRONMENT_SETUP.md
- System requirements
- Dependency installation
- Configuration steps
- Validation procedures
Operations Audit
File: OPERATIONS_AUDIT.md
- Training execution records
- Performance metrics
- Issue tracking
- Improvement recommendations
Scenario Mapping
File: SCENARIO_STAGE_MAPPING.md
- Real-world scenario mappings
- Use case correlations
- Practical application examples
- Integration patterns
Quick Start for Hermes
- Review
HERMES_LEARNING_PROMPT.mdfor detailed learning instructions - Analyze
training_schema.jsonto understand the stage structure - Study
ENVIRONMENT_SETUP.mdto understand environment requirements - Begin with Stage 1 (
stage1_foundation.json) and progress sequentially - For each stage, provide:
- Debug messages for issues found
- Suggestions for improvements
- Potential failure points
- Cross-stage dependency notes
- Use
SCENARIO_STAGE_MAPPING.mdto understand practical applications
Prerequisites
- AITBC blockchain node running and synchronized
- Python 3.13.5 with required dependencies
- Valid wallet accounts with sufficient AIT tokens
- Network connectivity to blockchain RPC endpoints
- Basic understanding of blockchain concepts
Validation
Each training stage includes validation criteria to ensure successful completion:
- Command execution success
- Expected state changes
- Transaction confirmations
- Balance verifications
- Health check validations
Troubleshooting
Hermes: When analyzing this section, identify:
- Common failure patterns across stages
- Missing troubleshooting steps
- Ambiguous error handling instructions
- Gaps in the troubleshooting coverage
For issues during training:
- Check
ENVIRONMENT_SETUP.mdfor configuration problems - Review
OPERATIONS_AUDIT.mdfor known issues and solutions - Verify blockchain node synchronization status
- Confirm wallet balances and permissions
- Check network connectivity to RPC endpoints
Related Resources
Quality Metrics
- Stage Coverage: 11 comprehensive training stages
- Schema Compliance: 100% adherence to training schema
- Documentation Completeness: All stages documented with examples
- Validation Coverage: Each stage includes success criteria
Last Updated: 2026-05-09 Version: 2.0 Maintained By: AITBC Development Team Changelog:
- v2.0 (2026-05-09): Added bidirectional cross-references between stages, JSON files, and scenarios
- v1.0 (2026-05-07): Initial version