Skip to content
#

capability-control

Star

Here are 5 public repositories matching this topic...

Language: All
Filter by language
All 5 HTML 2 JavaScript 1 Smalltalk 1 TypeScript 1
agent-tool-governance-gateway

putmanmodel / agent-tool-governance-gateway

Star 0

Runtime governance demo for tool-using AI agents: CDE detects deviation, Kingpin contracts or restores authority, and a thin gateway enforces the decision. Includes scoped leases, revocation, quarantine, human review, audit logs, and staged recovery.

agent middleware express policy governance ai-safety human-in-the-loop ai-agents fastapi guardrails tool-use llm runtime-governance human-in-the-loop-agents capability-control
  • Updated Sep 15, 2026
  • JavaScript

putmanmodel / kingpin-weak-signal-demo

Star 0

Runtime capability governance for AI agents. Kingpin separates what an agent notices or believes from what it is authorized to do, using scoped leases, human review, retry controls, prompt-injection handling, and deterministic audit behavior.

autonomous-agents ai-safety human-in-the-loop ai-agents ai-security runtime-security prompt-injection agentic-ai agent-security agentic-ai-security runtime-governance capability-governance ai-agents-security capability-control agent-security-tools
  • Updated Sep 14, 2026
  • TypeScript

bethediamond / ai-alignment-proxy

Star 0

Toy 5. An interactive proxy decay simulator showing how optimization pressure erodes the modeling capacity required to distinguish proxy from territory — producing self-reinforcing V(t) degradation that becomes progressively harder to correct. Companion simulation for The Depth Constraint — Series 2, Part 2.

javascript html machine-learning reinforcement-learning simulation optimization neural-networks dynamical-systems interactive-visualization ai-safety ai-research ai-alignment proxy-alignment agent-behavior reward-hacking alignment-research mesa-optimization goodharts-law capability-control value-learning
  • Updated May 28, 2026
  • HTML

bethediamond / ai-alignment-crossing

Star 0

Toy 3. An interactive model of the alignment phase ratio Φ = C / A_causal — the variable governing whether AI capability outpaces system-awareness before the crossing to stability can occur. Includes falsification test, oracle counterfactual, and point-of-no-return detection. Built to accompany The Alignment of Intelligence, Article 3: The Crossing

visualization javascript html reinforcement-learning simulation agent-based-modeling dynamical-systems feedback-systems ai-safety decision-theory complexity-theory phase-transition ai-alignment causal-modeling interactive-simulation falsifiability existential-risk alignment-research instrumental-convergence capability-control
  • Updated May 16, 2026
  • HTML

Add this topic to your repo

To associate your repository with the capability-control topic, visit your repo's landing page and select "manage topics."

Learn more