feat: add web4-governance plugin for trust-based AI governance

[dp-web4]

Integrates Web4 trust tensors with claude-flow's existing systems.

Features

• T3 Trust Tensors: Fractal 3D structure (Talent/Training/Temperament)

  • Each base dimension has subdimensions for fine-grained tracking
  • Composite formula: talent * 0.3 + training * 0.4 + temperament * 0.3

• Policy Evaluation: Rule-based tool authorization with trust thresholds

• Witnessing Chains: Bidirectional trust attestation between entities

• R6 Audit Chain: Hash-linked action logging for accountability

• Rate Limiting: Sliding window rate control per tool/category

Integration

Complements claude-flow's existing systems:

• Claims System: Extends authorization with trust scores
• AIDefence: Adds trust-weighted threat decisions
• Official Hooks Bridge: Policy enforcement at tool boundaries

Structure

• TypeScript integration layer (src/)
• Rust WASM module for performance-critical operations (wasm/)
• Policy presets for common governance scenarios

[dp-web4]

Just noticed the new @claude-flow/guidance control plane in PR #1065. These efforts appear complementary:

This PR (web4-governance)	Guidance Control Plane (#1065)
T3 Trust Tensors (Talent/Training/Temperament)	Generic trust accumulation/decay/tiers
Web4-specific trust semantics	Broad enforcement infrastructure
Plugin architecture	Core control plane

Potential integration: Our T3 tensors could serve as a pluggable trust scoring model within the guidance framework's trust primitive. The guidance control plane provides enforcement gates; we provide Web4 trust semantics.

Happy to discuss integration if there's interest in combining these approaches rather than running them as separate systems.