AI Agent Action Audit Log Worksheet for ERP Changes
If an AI agent can change ERP data, someone will eventually ask who asked for it, what ran and who said yes. This worksheet defines the audit fields and lets you keep a manual log during a pilot or review.
| Date | Requesting user | Tool called | ERP record | Action | Approver | Minutes to approve | Outcome | Note | |
|---|---|---|---|---|---|---|---|---|---|
Your real audit trail should be written automatically by the MCP server or gateway. Use this sheet to define the fields, run a pilot by hand, or reconcile exceptions during review.
Review my agent audit design
Optionally have the fields and retention of your audit trail reviewed.
Optional. The tool and every export work without this. No spam.
How to use the log
Add one row per tool call that matters: every write, every rejected request and every error, plus reads that touched sensitive data. Record the requesting user, the tool and the ERP record. For writes that need human approval, capture the approver and the minutes until they decided. The outcome column separates applied changes from declined and failed ones. Export to Excel for audit evidence and review the exceptions weekly during the first months of a pilot.
Fields every agent audit trail needs
Beyond what this sheet shows, a production log should hold a unique request ID, the model and prompt version, the exact tool arguments, the generated query and the ERP response status. Store the identity of the person behind the request, not just the service account, otherwise you cannot answer who asked. Keep the log append-only and out of the agent's own write reach.
- Request ID that ties prompt, tool call and ERP transaction together
- Requesting user and approver identities
- Tool arguments before and after validation
- Model and prompt version
- Before and after values for every update
Reading the example
The sample shows one approved date change after twelve minutes, one declined after forty-five and one failed requisition. The declined entry is the valuable one: it shows the approval step caught a conflict with the production plan. Average approval time also tells you whether the human gate is workable or becoming a bottleneck that users will route around.
Common mistakes
Logging only successful actions hides the failures auditors care about. Logging only the service account hides the user. Letting the agent itself write its own audit entries is weak because a faulty agent can also write a faulty log. Check retention against your regulatory needs and keep personal data in the log to the minimum.
Frequently Asked Questions
Which agent actions must be logged?
At minimum every write, every refusal, every error and every read of sensitive data. Many teams log all tool calls because storage is cheap and the log is the only way to reconstruct what happened.
Should the log be inside the ERP?
Use the ERP's own audit where available, but also keep an independent log at the MCP server or gateway. It captures the prompt and tool arguments, which the ERP does not see.
How long should we retain agent logs?
Follow your records retention policy and any regulation that applies to the underlying transactions. Financial changes often need years, while debug detail can be kept shorter.
Does this sheet replace automatic logging?
No. It defines the fields and supports pilots and reviews. Production logging should be automatic and tamper resistant.
Netray's ERP agents log the question, the generated query and approvals so changes are traceable. If you are designing the audit trail for your own build, we can review the fields.
Related Tools
MCP Tool Catalog and Risk Worksheet
Register each MCP tool your agent exposes, the ERP object it touches and a computed risk score, then export to Excel.
Developer ToolsAI Agent ERP Access Request Form Template
A fill-in form that records which ERP roles, objects and limits an AI agent service account needs, with approval signatures.
Compliance & ContractsAI Agent Incident Response Runbook Template
A fill-in incident report for AI agent mistakes in the ERP: timeline, containment, root cause, data affected and corrective actions.
AI Agents & AutomationAI Agent Evaluation Test Case Worksheet
Log evaluation questions for an ERP agent with expected answers, the tool it should call and scored results, with a computed pass rate.
More Compliance & Contracts tools
See all- SAP SoD Residual Risk Score Calculator
- SAP Firefighter ID Usage Review Log
- IBM i (AS/400) User Profile Access Review Worksheet
- SyteLine Security and User Access Audit Checklist
- Infor XA (MAPICS) User Access Review Worksheet
- China ERP Data Security and Compliance Assessment
- ERP Vendor Data Processing Agreement Outline
- GDPR ERP Data Migration Checklist
- EU Machinery Regulation 2023/1230 Readiness Checklist