EN
Agent Governance
Budgets, approvals, permissions, audit logs β the controls that keep agents safe.
- Agent Governance: control and management of AI agents in productionβ β βA practical production framework for controlling AI agents: access, limits, approval, audit logs, kill switch, and rollback.
- RBAC For AI Agents: Role-Based Access Control Without Excess Privilegesβ β βPractical RBAC for AI agents in production: roles, tenant scope, default deny, approval for write actions, and audit trail.
- Budget Controls For AI Agents: How To Limit Runtime Spendβ β βPractical budget controls in production: max_steps, max_seconds, max_tool_calls, max_usd, stop reasons, audit logs, and alerting.
- Human Approval For AI Agents: How To Safely Control Write Actionsβ β βPractical approval flow in production: approval_required, TTL, stop reasons, approval token, and audit trail for write actions.
- Allowlist vs Blocklist For AI Agents: Why Default Deny Is Saferβ β βPractical approach to tool access in production: default-deny allowlist, incident blocklist, stop reasons, and audit logs.
- Step Limits for AI Agents: how to stop loops before an incidentβ β βStep limits in production: how to stop loops, return stop reasons, and keep run execution under control.
- Kill Switch for AI Agents: how to emergency-stop actions without a releaseβ β βPractical kill switch in production: global/per-tenant stop, writes-disable mode, stop reasons, audit trail, and a short runbook.
- Rate Limiting for AI Agents: how to contain request spikes and retry stormsβ β βPractical rate limiting in production: per-user/per-tenant/global limits, burst control, retry_after, backoff, audit logs, and alerting.
- Agent Versioning for AI Agents: how to safely release prompt, tools, and policyβ β βPractical agent versioning in production: version manifest, contract checks, canary rollout, rollback, and run reproducibility.
- Rollback Strategies for AI Agents: how to safely roll back a releaseβ β βPractical rollback in production: stop reasons, traffic switch to stable version, canary gates, audit logs, and a tested runbook.
- Audit Logs for AI Agents: how to reconstruct decision chains in productionβ β βPractical audit trail in production: policy decisions, stop reasons, actor/scope, redaction, immutable storage, and fast incident investigation.
- Multi-Agent Governance: coordination, permissions, and escalation controlβ β βPractical multi-agent governance in production: role ownership, handoff limits, shared budgets, approval gates, stop reasons, and audit trail.