Runtime prompt injection defense for OpenClaw. Protects AI agents from attacks via inputs, tool results, and memory poisoning.
Stars
6
Forks
1
Watchers
6
Open Issues
0
Overall repository health assessment
^0.32.0^4.1.2^16.4.5^9.0.5^1.0.2^3.23.8^29.5.12^20.14.0^29.7.0^29.2.0^4.19.0^5.5.013
commits
docs: split README into crash course + dedicated call for testers
192d6e1View on GitHubfeat: Agent Smith v2 - Capability-based security + Secret vault
31f27a1View on GitHubdocs: lead with uncensored-LLM pitch, expand For Uncensored LLMs section
086c319View on GitHubdocs: state that docs and code are in English for contributors
798ee92View on GitHubfeat: Agent Smith v0.1 – core detector, CLI, 20 attacks + 20 legitimate corpus, 43 tests
fb0b971View on GitHub