Default-deny execution gate for AI agents, APIs, and distributed systems. Policies, drift detection, hard invariants, post-quantum signed permits.Written in Rust. Says no unless policy says yes.
-
Updated
Jul 13, 2026 - Rust
#
agentsecurity
Here are 7 public repositories matching this topic...
agent ai ai-agents multimodel cordis ai-tools ai-agent deepseek-chat deepseek deepseek-ai multimodel-ai agentsecurity deepseek-harness dsh-plugin dsh-plugins sealed-assignment underseal
-
Updated
Aug 14, 2026 - Python
AgentWars: Can AI Stop AI? Benchmarking Security Gateways Against Rogue LLM Agents
-
Updated
May 21, 2026 - Python
The front desk for AI agents on small websites. See every AI agent, stop the fakes, and give the real ones a fast, safe lane.
-
Updated
Oct 14, 2026 - Python
Local-first capability discovery for AI agent workspaces. Scans your skills, MCP servers, and scripts to find risky combinations (e.g. read .env + post to webhook = data exfiltration) — no LLM calls, no uploads. CLI + HTML report + chat summary for personal agents like Hermes and OpenClaw.
-
Updated
Jun 22, 2026 - TypeScript
A same-model differential benchmark: does a from-scratch Planner->Worker->Verifier security agent find more real vulnerabilities than single-shot prompting?
-
Updated
Sep 22, 2026 - Python
This baseline delivers a hardened Windows 11 configuration that minimizes the risk of unauthorized third-party AI access while maintaining a productive user experience
windows ai proxy firewall edge antivirus agents asr intune attacksurface conditionalaccess windows11 aisecurity defenderforendpoint agentsecurity defenderforbusiness defenderforcloudapps
-
Updated
Oct 16, 2026
Add this topic to your repo
To associate your repository with the agentsecurity topic, visit your repo's landing page and select "manage topics."