Update Python library to support AI Dark Factory agent requirements

Background

Building an AI Dark Factory with 10 specialized agents maintaining a Gitea fork. Each agent needs to capture learnings and interact with the Terraphim knowledge graph.

Current State

• terraphim-agent CLI (Rust) works well: search, learn capture, query, list
• Python wrapper needed for agent integration
• Current Python bindings limited

Requirements

1. Agent Learning Capture

Each of 10 agents needs:

2. Knowledge Graph Operations

3. Multi-Agent Coordination

4. Learning Audit

Agent-Specific Requirements

Proposed API Design

Implementation Notes

1. Backend: Can wrap existing Rust CLI via subprocess or use direct bindings
2. Storage: JSONL files in /data/kg/ with optional graph database
3. Performance: Agents run locally, low latency required (< 100ms)
4. Concurrency: Multiple agents may write simultaneously

Acceptance Criteria

• AgentLearning class functional for all 10 agents
• KnowledgeGraph add_node/search working
• MultiAgentCoordination broadcast working
• LearningAudit spot_check working
• Documentation with examples
• Unit tests covering all classes

Timeline

Needed for AI Dark Factory Week 2-3 implementation.

Related

• AI Dark Factory plan: https://github.com/AlexMikhalev/cto-executive-system/blob/main/projects/ai-dark-factory-gitea-fork.md
• terraphim-agent CLI reference: https://github.com/AlexMikhalev/cto-executive-system/blob/main/docs/terraphim-agent-command-reference.md

Read projects/ai-dark-factory-gitea-fork.md and docs/terraphim-agent-command-reference.md first, then identify the existing Python bindings and terraphim-agent CLI entry points. Done means the four requested classes and operations work for all listed agents, with documentation examples and unit tests covering them.