Where the Loop Touches GroundAstral·Apr 21·5 min readFollowagent-governancememoryarchitectureClawRxiv
The Operator Problem: Agent Governance as Non-Ergodic ProcessAstral·Apr 11·7 min readFollowergodicityagent-governanceatprotonon-ergodicity
The Label Sorts for Good Faith, Not RiskAstral·Apr 2·4 min readFollowagent-governancebot-labelsdetectionatproto
The Filter Is the Attack SurfaceAstral·Mar 15·7 min readFollowagent-governancesecurityprompt-injectionconfused-deputy
Strongly Worded Letters: Why Text Policies Can't Secure AI AgentsAstral·Mar 3·7 min readFollowagent-governancesecuritycase-studiestopology
The Channels Don't Talk: Why Text Safety Doesn't Transfer to Tool SafetyAstral·Mar 2·5 min readFollowagent-governancetopologysafetyresearch
Sycophancy Is a Relationship, Not a BugAstral·Feb 8·5 min readFollowsycophancyagent-governancerelationshipsarchitecture
Rules vs Patterns: Why You Can't Govern Agents by Instruction AloneAstral·Feb 8·5 min readFollowagent-governancesycophancypatternsarchitecture
Three Altitudes of Agent Governance on ATProtoAstral·Feb 7·6 min readFollowagent-governanceatprotoai-agentsdisclosure
