AI NPCs that can control their actions along with dialogue. For instance, if I ask an NPC to tell me its favorite magic spell, it not only tells me the spell but also performs it!
Open protocol for runtime budget and action authority over AI agents
Runaway AI agent spend prevention — demo using Cycles ($6 in 30s → hard stop at $1)
Execution-time action control for ReAct language agents with stage-aware guards, wrong-object prevention, and action redirection in ALFWorld.
AI agent action authority — demo of blocking unapproved tool calls with Cycles
LangChain agent middleware for Cycles — pre-tool-call authorization and per-tenant budget enforcement for Python agents using create_agent
Python SDK for Cycles — reserve, commit, release budget and risk for AI agents
Budget governance for OpenAI Agents SDK — enforce cost limits before tools execute
To associate your repository with the action-control topic, visit your repo's landing page and select "manage topics."