Show HN: open-weight LLM with internally flippable answer/stop decision
Show HN: Eliminating text looping/latent collapse in LLM activation steering
Show HN: Eliminating text looping/latent collapse in LLM activation steering
A local-first tool for red-teaming LLM reasoning and agent actions, providing honest scoring. It helps security researchers and developers evaluate and improve the robustness of LLM-based systems.
A local-first tool for red-teaming LLM reasoning and agent actions, providing honest scoring. It helps security researchers and developers evaluate and improve the robustness of LLM-based systems.
ReasonGate is an open-source tool that acts as a gate to block prompt injection attacks on LLM applications. It provides explainable decisions, helping developers secure their AI systems against malicious inputs.
ReasonGate is an open-source tool that acts as a gate to block prompt injection attacks on LLM applications. It provides explainable decisions, helping developers secure their AI systems against malicious inputs.
Governed Agent is a framework where an LLM interprets user requests, but actual actions are controlled by predefined rules and code. It solves the problem of ensuring LLM outputs comply with business logic and safety constraints, useful for developers building reliable AI agents.
Governed Agent is a framework where an LLM interprets user requests, but actual actions are controlled by predefined rules and code. It solves the problem of ensuring LLM outputs comply with business logic and safety constraints, useful for developers building reliable AI agents.
Descles provides response-level control over LLM-generated messages, routing outputs through human-in-the-loop approval flows before they are sent. It targets teams using LLMs in customer-facing messaging who need oversight of automated responses.
Descles provides response-level control over LLM-generated messages, routing outputs through human-in-the-loop approval flows before they are sent. It targets teams using LLMs in customer-facing messaging who need oversight of automated responses.