Application security review
Design and code reviews, threat modeling, and a secure SDLC that developers actually follow.
應用程式資安審查:設計與程式碼審查、威脅建模,以及開發團隊真的會照著走的 secure SDLC。
Security consulting
資安三昧 · 資安顧問
Calm, clear security for applications, AWS, and AI agents.
替應用程式、AWS 與 AI agent 把資安做得穩、看得清。
Design and code reviews, threat modeling, and a secure SDLC that developers actually follow.
應用程式資安審查:設計與程式碼審查、威脅建模,以及開發團隊真的會照著走的 secure SDLC。
AWS posture against the CIS benchmark, fixes for misconfigurations, and drift monitoring that shows who changed what.
雲端資安檢測:依 CIS benchmark 檢查 AWS、修正錯誤設定,並監控設定漂移、找出誰改了什麼。
Guardrails for LLM applications and AI coding agents, risk assessment, and AI governance.
AI 與 agent 資安治理:LLM 應用與 AI coding agent 的護欄、風險評估與 AI 治理。
vCISO support, a security program built from zero, and controls mapped to frameworks and regulations.
資安計畫與 GRC:vCISO、從零建置資安計畫,把控制措施對應到框架與法規。
A read-only CIS drift monitor for AWS, built for the Daytona HackSprint. It takes a Prowler baseline, diffs resource snapshots, uses CloudTrail to find who changed what, and explains every regression with an LLM. Tested against a deliberately misconfigured AWS lab.
為 Daytona HackSprint 打造的唯讀 AWS CIS 漂移監控:Prowler 基準掃描、資源快照比對、用 CloudTrail 找出誰改了什麼,每個退步都由 LLM 說明原因。在刻意設錯的 AWS 靶場上實測過。
A multi-agent security review assistant in Slack. One security lead routes each request to specialist reviewers (AWS, SaaS, PCI on Snowflake, AppSec, vendor risk, audit evidence, architecture) and returns a single report.
Slack 裡的資安 multi-agent 審查助手:由一個 security lead 把請求分派給 AWS、SaaS、Snowflake PCI、AppSec、供應商風險、稽核證據、架構等專家 reviewer,最後彙整成一份報告。
Verification gates before every commit, hooks that catch deleted tests and weakened assertions, an allowlisted DevOps MCP server, and a blameless post-mortem of a real file-deletion incident.
AI coding agent 的護欄:每次 commit 前的驗證關卡、抓出被刪測試與被弱化 assertion 的 hook、只開白名單動作的 DevOps MCP server,以及一次真實刪檔事故的無究責檢討。
Security Engineer · 資安工程師