fableforge-agent 0.1.0

Self-improving AI agent framework with behavioral observability and trust protocols

╔═══════════════════════════════════════════════════════════════╗ ║ ███████╗ █████╗ ██████╗ ██╗ ███████╗ ║ ║ ██╔════╝██╔══██╗██╔══██╗██║ ██╔════╝ ║ ║ █████╗ ███████║██████╔╝██║ █████╗ ║ ║ ██╔══╝ ██╔══██║██╔══██╗██║ ██╔══╝ ║ ║ ██║ ██║ ██║██████╔╝███████╗███████╗ ║ ║ ╚═╝ ╚═╝ ╚═╝╚═════╝ ╚══════╝╚══════╝ ║ ║ ███████╗ ██████╗ ██████╗ ██████╗ ███████╗ ║ ║ ██╔════╝██╔════╝ ██╔══██╗██╔════╝ ██╔════╝ ║ ║ █████╗ ██║ ███╗██████╔╝██║ ███╗█████╗ ║ ║ ██╔══╝ ██║ ██║██╔══██╗██║ ██║██╔══╝ ║ ║ ██║ ╚██████╔╝██║ ██║╚██████╔╝███████╗ ║ ║ ╚═╝ ╚═════╝ ╚═╝ ╚═╝ ╚═════╝ ╚══════╝ ║ ║ ║ ║ Self-improving AI agent framework with behavioral observability ║ ║ and trust protocols. Built on Hermes Agent by NousResearch. ║ ║ ║ ║ Author: KingLabsA ║ ║ • GitHub: https://github.com/KingLabsA ║ ║ • Twitter: https://x.com/KingLabsA ║ ║ • Email: [e-mail] ║ ╚═══════════════════════════════════════════════════════════════╝

---

FableForge Agent is a self-improving AI agent framework built on top of Hermes Agent by NousResearch. It extends Hermes with three additional tiers of capability:

• Tier 3: Agent Control Plane — Sandboxed execution, audit trails, incident response, OAuth gate, and credential broker.
• Tier 4: Behavioral Observability — Real-time behavior observation, semantic firewall, safety rules engine, and behavior reporting.
• Tier 4: Trust Protocols — Bidirectional agent-to-agent trust, identity registry, permission framework, and cryptographic signing.

Quickstart

pip install fableforge-agent
fableforge --help

Or run with specific subcommands:

fableforge                          # Run the agent (passes through to Hermes CLI)
fableforge control-plane --port 8000  # Start the control plane API server
fableforge audit --export report.json # Export audit trail
fableforge trust register --agent-id agent1 --pubkey ./key.pub  # Trust registry
fableforge observe                   # Start behavioral observability dashboard

Architecture

┌─────────────────────────────────────────────────────┐
│                   FableForge Agent                    │
├─────────────────────────────────────────────────────┤
│  Tier 4:  Behavioral Observability + Trust Protocols │
│  ┌────────────────────┐  ┌────────────────────────┐  │
│  │ BehavioralObservability│  │   TrustProtocol       │  │
│  │ SemanticFirewall     │  │   IdentityRegistry    │  │
│  │ BehaviorReport       │  │   PermissionFramework │  │
│  └────────────────────┘  └────────────────────────┘  │
├─────────────────────────────────────────────────────┤
│  Tier 3:  Agent Control Plane                        │
│  ┌────────────────────┐  ┌────────────────────────┐  │
│  │ Sandbox (execution)│  │   AuditTrail           │  │
│  │ IncidentResponse   │  │   OAuthGate            │  │
│  │ CredentialBroker   │  │   SandboxPolicy        │  │
│  └────────────────────┘  └────────────────────────┘  │
├─────────────────────────────────────────────────────┤
│  Tiers 1-2:  Hermes Agent Foundation                 │
│  ┌──────────────────────────────────────────────┐   │
│  │ AIAgent · ToolRegistry · ConversationLoop    │   │
│  │ Model Providers · Tool Calling · Memory      │   │
│  └──────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────┘

Foundation

FableForge Agent is built on Hermes Agent by NousResearch. All credit for the underlying agent framework, tool calling, conversation loop, and model provider integration belongs to the NousResearch team.

License

Apache 2.0 — see LICENSE for details.