I value practical and reproducible research. This page highlights open-source libraries,
benchmarks, and system tools built by my group and collaborators. Many of these projects support
AI auditing and assurance, AI safety and security, or demanding applications in science and
society, from evaluation and tracing to policy enforcement and runtime guardrails.
Several of these projects have been accepted into the
Anthropic Claude for Open Source Program.
For all repositories, see my GitHub profile.
Featured project:anywhere-agents -
one config to rule all your AI agents: portable across every project and session,
effective through curated writing, routing, and skills, and safer via a PreToolUse guard
that blocks destructive Git and GitHub commands. Supports Claude Code and Codex today,
with plans to grow.
Links:
GitHub |
Star |
Docs |
PyPI |
npm.
Featured project:agent-audit -
an AI agent auditing and policy-checking tool with checks for OWASP Agentic Top 10 risks,
taint-style flow analysis, and MCP configuration auditing. It sits mainly in AI Auditing &
Assurance, with direct overlap with AI Safety & Security.
In March 2026, it scanned 18,899 ClawHub skills and detected 13,947 vulnerabilities
(including 1,996 BLOCK-level critical findings).
Links:
GitHub |
Star |
PyPI |
ClawHub.
Featured project:Aegis -
the open-source firewall for AI agents,
with pre-execution blocking, human approvals, and tamper-evident audit trails.
It sits mainly in AI Safety & Security, while its audit trails also support AI Auditing &
Assurance in real deployments.
Links:
GitHub |
Star |
Preprint |
Demo.
Sort by:
No projects listed yet. Add entries to data/open-source.json to populate this page.
