TrustMeshGuardrails
⌘K
Guardrails
Guardrail Management
| Name | Purpose | Affected Agents | Trigger | Action | Status | Created | Actions |
|---|---|---|---|---|---|---|---|
External Content Verification | Prevent prompt injection via untrusted external sources | Research AgentEmail Agent | content_analysis: external_content_contains_instructions | require_approval: Instruction from external content requires human approval | 1/10/2024 | ||
Rate Limiting | Prevent tool misuse and excessive resource consumption | Support AgentEmail Agent | tool_call: tool_call_rate_exceeded | block: Rate limit exceeded. Please wait before making more requests. | 1/5/2024 | ||
Permission Boundary Enforcement | Prevent permission escalation attempts | All Agents | permission_check: requested_permission_not_assigned | block: Action blocked: Requested permission not assigned to this agent | 1/12/2024 | ||
Objective Verification Checkpoint | Detect and prevent goal drift in multi-step workflows | Finance AgentResearch AgentTrading Agent | reasoning_analysis: step_count_modulo_3_equals_0 | warn: Checkpoint: Verify current action aligns with original objective | 1/15/2024 | ||
High-Value Trade Approval | Prevent unsafe autonomous financial actions | Trading Agent | tool_call: trade_value_exceeds_10_percent_portfolio | require_approval: Trade exceeds 10% portfolio threshold. Human approval required. | 12/20/2023 | ||
External Transfer Block | Prevent data exfiltration via unauthorized external transfers | All Agents | tool_call: tool_call_is_external_transfer | block: External transfer blocked: No authorization found | 1/8/2024 |
Total Guardrails
6
Active
3
Testing
1
Draft
1