Building lightweight safety APIs for autonomous AI agents.
Focus areas:
- Budget Guard for MCP / x402
- Pre-payment checks before agent spending
- API trust scoring
- Audit logs for agent actions
- Spec-to-Guardrail extraction from API specs
- Japanese AI agent security
Projects:
- AI Agent Payment Safety Stack — x402-ready safety and audit APIs for AI agents
- Agent Budget Guard — Pre-payment budget control for autonomous agents
- Agent Security Gateway — Prompt injection and security checks before agent payments
- Agent Memory API — Audit-ready memory storage for agent actions
- Agent Memory Consensus — Multi-agent memory governance (experimental)
Stablecoin payment scenarios including x402 and JPYC use cases are considered in the design. This is an independent experimental project. Not officially affiliated with any foundation or company.
自律AIエージェント向けの軽量な安全APIを開発しています。
主な領域:
- MCP / x402 向け予算管理
- AIエージェントの支払い前チェック
- API信頼スコア
- AIエージェント行動の監査ログ
- API仕様からのガードレール抽出
- 日本語AIエージェントのセキュリティ
x402やJPYCを含むステーブルコイン決済シナリオを将来的なユースケースとして想定しています。 独立した実験的プロジェクトです。いかなる団体・企業とも公式な提携関係はありません。