liveBook · Manning

Audit trail

As explained in Vibe Engineering

An audit trail is a chronological record of actions and changes that supports accountability and later inspection.

An audit trail is a chronological record of actions and changes used to inspect decisions, assign responsibility, and satisfy governance demands. In agent-driven development, security may require every agent action to be logged, while compliance may require evidence that a human reviewed generated code. Informal trust is no longer enough: agents need explicit, inspectable records. section 5.5

Build auditability around visible work products rather than relying only on chat histories:

  • Preserve planning decisions, implementation context, test results, and documentation as artifacts that can be inspected at workflow boundaries. section 5.6.5
  • Insert human checkpoints before implementation, before testing, and before merge; approval should identify what was reviewed rather than merely record repeated clicks. section 5.5 section 5.6.5
  • Treat artifacts as evidence, not as complete context: handoffs can omit assumptions, rejected alternatives, and discovered edge cases. section 5.6.5
  • Prefer sanctioned tools with proper audit trails over blanket bans, which can drive usage into shadow workflows that cannot be seen or secured. section 10.5

A four-agent handoff

A planning agent produces a design document, an implementation agent writes code, and later agents inspect tests and documentation. If a failure appears, the artifacts show what the planner decided, what context implementation received, and what verification actually covered. The trail improves forensics, but reviewers must still check whether important assumptions disappeared at the handoff. section 5.6.5

How Vibe Engineering covers it

How deep it goes
Explains logging, artifact checkpoints, and forensic value, but stops short of specifying a concrete logging implementation.
Examples
Uses agent actions, design documents, test results, artifact timelines, approval checkpoints, and sanctioned pipelines as evidence.
What you can do after
Operate agent workflows with inspectable artifacts, explicit review checkpoints, and evidence for debugging or compliance.
How it connects
Connects audit trails with governance, compliance, human oversight, software agents, artifacts, verification, and shadow AI.

Also explained in

Compare how 5 books explain audit trail →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Vibe Engineering. The section links point to the parts of the book it draws on.