I build local-first tools for AI-agent operators, blockchain experiments, and systems that need to stay auditable under pressure. I like automation, but only after the rollback path, the logs, and the trust boundary are clear.
- Shipping Agent Safety Kit: local-first safety rails for AI-agent operators.
- Exploring agent continuity, reversible file operations, skill-intake gates, and memory hygiene.
- Still close to blockchain: protocols, wallets, audit trails, incentives, and trust-minimized systems.
agent safety rollback-first tools, checkpoints, local reports
systems CLIs, automation, validators, operational dashboards
blockchain smart contracts, protocol experiments, wallet/payment flows
product small tools that can become real businesses
- Prefer boring reliability over magical demos.
- Keep private data local by default.
- Make automation auditable before making it autonomous.
- Ship small, test the edges, then compound.
- Agent Safety Kit을 만들고 있습니다. AI 에이전트가 파일을 수정하거나 도구를 설치하거나 긴 작업을 이어갈 때 필요한 로컬 우선 안전장치입니다.
- 에이전트 연속성, 되돌릴 수 있는 파일 작업, 스킬 설치 전 검토, 메모리 위생 같은 운영 안정성 문제를 다룹니다.
- 블록체인 쪽도 계속 보고 있습니다. 프로토콜, 지갑, 감사 추적, 인센티브, 신뢰 최소화 시스템에 관심이 있습니다.
에이전트 안전 롤백 우선 도구, 체크포인트, 로컬 리포트
시스템 CLI, 자동화, 검증기, 운영 대시보드
블록체인 스마트컨트랙트, 프로토콜 실험, 지갑/결제 흐름
제품 작지만 실제 비즈니스가 될 수 있는 도구
- 화려한 데모보다 오래 버티는 안정성을 선호합니다.
- 기본값은 로컬 우선, 프라이버시 우선입니다.
- 자동화하기 전에 먼저 감사 가능하게 만듭니다.
- 작게 만들고, 경계 조건을 확인하고, 누적해서 키웁니다.
- 正在构建 Agent Safety Kit:面向 AI 代理操作者的本地优先安全工具。
- 关注代理连续性、可回滚文件操作、技能安装前审查、记忆卫生和可审计工作流。
- 也持续关注区块链:协议、钱包、审计轨迹、激励机制和最小信任系统。
代理安全 回滚优先工具、检查点、本地报告
系统工程 CLI、自动化、验证器、运维仪表盘
区块链 智能合约、协议实验、钱包/支付流程
产品 小而清晰、可以发展成业务的工具
- 比起炫酷演示,更重视可靠性。
- 默认本地优先,默认保护隐私。
- 在追求自治之前,先保证可审计。
- 小步发布,测试边界,然后持续复利。
| Project | What it is |
|---|---|
| Agent Safety Kit | Local-first safety rails for people running AI agents that edit files, install tools, or continue work across sessions. |
| Agent Safety Kit Pro | Private pro analyzers for context pulse, taskflow audit, memory hygiene, dashboards, and CI checks. |
| Blockchain experiments | Protocol and payment experiments around auditability, incentives, and operational trust. |
local-first · agent safety · blockchain systems · developer tools · quietly intense nerd energy
