diff --git a/README-zh.md b/README-zh.md
index 5c00d9d..849364e 100644
--- a/README-zh.md
+++ b/README-zh.md
@@ -225,6 +225,7 @@
 - [shellgate](./plugins/shellgate)
 - [web-security-guard](./plugins/web-security-guard)
 - [agent-safety-preflight](./plugins/agent-safety-preflight)
+- [sofagent](https://github.com/KongFangXun/sofagent) - 外部：提交时审计 AI 编码智能体的 FDE Harness——24 条审计规则扫描 git diff（密钥泄露、越界修改、未读先改、提交信息注入），写入 HMAC 签名的防篡改审计史，附 MCP 治理工具。MIT。
 
 
 ### MCP 服务器
diff --git a/README.md b/README.md
index 0670365..1fe9626 100644
--- a/README.md
+++ b/README.md
@@ -361,6 +361,7 @@ Install or disable them dynamically with the `/plugin` command — enabling you
 - [openclaw-security](https://github.com/AtlasPA/openclaw-security) - 11-tool security suite for agent workspaces: integrity scanning, secret detection, permission auditing, network DLP, supply-chain checks, injection defense, credential lifecycle, compliance enforcement, audit trails, cryptographic signing, and incident response.
 - [privacy-mask](https://github.com/fullstackcrew-alpha/privacy-mask) - Detect and redact PII, API keys, and secrets in screenshots via OCR + regex before images reach Claude. 47 detection rules, Tesseract/RapidOCR support, UserPromptSubmit hook integration. 100% offline.
 - [skillfortify](https://github.com/qualixar/skillfortify) - External: Formal verification scanner for AI agent skills (including Claude Code skills + MCP servers). 100% precision on 540-skill benchmark. ASBOM generation. arXiv:2603.00195.
+- [sofagent](https://github.com/KongFangXun/sofagent) - External: FDE harness that audits AI coding agents at commit time — scans git diffs with 24 audit rules (secret leaks, out-of-scope edits, blind modifications without prior reads, commit-message prompt injection), writes HMAC-signed tamper-evident audit history, and ships MCP governance tools. MIT.
 - [shellgate](./plugins/shellgate)
 - [huiyu-safe-ai](https://github.com/huiyu9144/huiyu-safe-ai) - Lightweight AI security guard for install/download commands. Blocks 68+ malicious packages, verifies identity, scans code in <1s. Built from a real supply chain attack.
 - [web-security-guard](./plugins/web-security-guard)