> ## Documentation Index > Fetch the complete documentation index at: https://splinter.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Control Layer Overview > Runtime Controls for AI agents The Control Layer provides **Runtime Controls** that prevent AI agents from causing harm. Without these controls, agents can spend unlimited money, loop forever, and break production systems. ## Why Control Matters Without proper control mechanisms, AI agents can cause severe operational and financial damage. Agents may accumulate massive API costs by making unlimited calls to expensive services, enter infinite loops that waste computational resources, overwhelm external APIs with excessive requests leading to rate limit violations, make unauthorized changes to critical systems, or consume unbounded memory until systems crash. These scenarios can transform helpful automation into costly failures that require immediate intervention and system shutdowns. The Control Layer prevents: | Problem | Control | What it does | | ----------------------- | -------------------- | ---------------------- | | Unlimited spending | ExecutionLimits | Hard budget cap | | API rate exceeded | RateLimiter | Calls per minute | | Cascading failures | CircuitBreaker | Stop after N failures | | Infinite loops | LoopDetection | Detect and break loops | | Unauthorized actions | ToolAccessController | Per-agent permissions | | Flip-flopping decisions | DecisionEnforcer | Lock decisions | | Transient failures | RetryStrategy | Retry with backoff | | Custom violations | RulesEngine | Your own rules | | Memory bloat | MemoryStore | Capped storage | ## Quick Start ```python theme={null} from splinter import Splinter # Basic limits - that's it! s = Splinter( openai_key="sk-...", max_budget=5.0, # Stop at $5 max_steps=50, # Stop after 50 calls ) result = await s.run("agent", "Do the task") # Guaranteed to stop at $5 or 50 calls, whichever comes first ``` ## Control Objects Budget, step count, and time limits Calls per minute per agent or tool Stop after repeated failures Detect and break infinite loops Which agent can use which tools Lock decisions to prevent flip-flopping Retry failed calls with exponential backoff Custom BLOCK/WARN/LOG rules Capped memory with auto-eviction ## How Controls are Applied ```mermaid theme={null} flowchart TB Request[Agent Request] RateLimit{Rate Limiter} RateBlock[❌ Too fast? Block] Circuit{Circuit Breaker} CircuitBlock[❌ Circuit open? Block] ToolAccess{Tool Access} ToolBlock[❌ Not allowed? Block] Rules{Rules Engine} RulesBlock[❌ Rule violation? Block] ExecLimits{Execution Limits} BudgetBlock[❌ Over budget? Block] LLM[LLM Provider] LoopDetect{Loop Detection} LoopBreak[❌ Loop detected? Break] Response[Agent Response] Request --> RateLimit RateLimit -->|✓ Pass| Circuit RateLimit -.->|Block| RateBlock Circuit -->|✓ Pass| ToolAccess Circuit -.->|Block| CircuitBlock ToolAccess -->|✓ Pass| Rules ToolAccess -.->|Block| ToolBlock Rules -->|✓ Pass| ExecLimits Rules -.->|Block| RulesBlock ExecLimits -->|✓ Pass| LLM ExecLimits -.->|Block| BudgetBlock LLM --> LoopDetect LoopDetect -->|✓ Pass| Response LoopDetect -.->|Break| LoopBreak style Request fill:#DBEAFE,stroke:#3B82F6,stroke-width:2px,color:#000 style RateLimit fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style Circuit fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style ToolAccess fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style Rules fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style ExecLimits fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style LLM fill:#D1FAE5,stroke:#10B981,stroke-width:2px,color:#000 style LoopDetect fill:#FEE2E2,stroke:#6c0404,stroke-width:2px,color:#000 style Response fill:#DBEAFE,stroke:#3B82F6,stroke-width:2px,color:#000 style RateBlock fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 style CircuitBlock fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 style ToolBlock fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 style RulesBlock fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 style BudgetBlock fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 style LoopBreak fill:#FEE2E2,stroke:#DC2626,stroke-width:2px,color:#000 ``` ## Combining Controls Controls compose naturally: ```python theme={null} from splinter import Splinter from splinter.types import ExecutionLimits, LoopDetectionConfig from splinter.control import RateLimiter, CircuitBreaker, ToolAccessController # Basic limits s = Splinter(openai_key="sk-...", max_budget=10.0, max_steps=100) # Add rate limiting limiter = RateLimiter() limiter.set_agent_limit("researcher", calls=20, window_seconds=60) # Add circuit breaker breaker = CircuitBreaker( breaker_id="openai", failure_threshold=5, timeout_seconds=60, ) # Add tool access control tool_ctrl = ToolAccessController() tool_ctrl.set_allowed_tools("researcher", ["web_search", "read_file"]) tool_ctrl.set_allowed_tools("writer", ["write_file"]) # All controls work together ``` ## Best Practices Even for testing. A runaway agent can cost thousands. Default to \$5-10 for development. If an external API is down, stop hammering it. Give it time to recover. Agents that run without human oversight need loop detection. They will get stuck. Don't give all agents access to all tools. Principle of least privilege.