Observe · Platform

Immutable Audit Trail

Every agent action, logged in real time and queryable. Your compliance evidence, always ready.

Prefactor captures an immutable record of every agent action, span, policy decision, approval event, and quality or risk signal as it happens. When regulators or auditors ask how your agents are evaluated and controlled, the answer is a live system with searchable, exportable evidence.

Audit log Illustrative
policy.evaluate support-agent logged
approval.granted billing-recon logged
pii.redacted triage-bot logged
TL;DR

Every agent run becomes an instance, and every step inside it — an LLM call, a tool invocation, a message — is recorded as a span with its input, output, duration, and outcome. Instances are immutable once written, giving reviewers a real-time, queryable record of exactly what every agent did.

What the audit trail captures

Every agent action, span, policy evaluation, enforcement decision, and approval event is recorded with full context — which agent, which policy or quality signal triggered the event, what decision was made, and when — the moment it happens. Records are tamper-evident, so an entry can't be quietly edited after the fact; the trail a regulator or an internal investigation sees is the same one Prefactor generated at the time.

Because every event links back to the registered agent that produced it, an auditor can trace a single production incident all the way from the triggering action back to the policy or quality signal that flagged it.

Querying and searching

The audit trail is full-text searchable and filterable by agent, team, time range, action type, policy, or enforcement decision, so investigating an incident means searching for what happened rather than scrolling through raw logs. The same search interface generates compliance reports on demand — a filtered view across a time range and a set of agents, ready to export.

Compliance and export

Exports are available as PDF, CSV, or JSON, and integrate with the SIEM and GRC tools a security or compliance team already uses, so audit evidence doesn't require a separate manual process to leave Prefactor. The trail maps to the specific compliance frameworks Prefactor supports — EU AI Act, GDPR, HIPAA, SOC 2, ISO 27001, ISO 42001, NIST AI RMF, and OWASP LLM Top 10 — with per-framework guidance at /compliance covering exactly which records satisfy which provisions.

Frequently asked questions

What does the audit trail capture?
Every agent run (an instance) and every step inside it (a span) — an LLM call, a tool invocation, a message sent or received — along with its inputs, outputs, duration, success or failure, agent version, environment, and risk classification where applicable.
Can audit records be edited or deleted?
No. Instances and spans are immutable once written — if a span's payload doesn't match its declared schema, the mismatch is recorded and flagged for review rather than silently corrected, so the trail a reviewer sees is the one Prefactor generated at the time.
How do compliance teams use the audit trail?
A reviewer filters to a specific agent's instances, narrows to the relevant environment and time window, and opens the instances that reached a risk level worth attention — turning an audit from a log-scrolling exercise into a targeted review.

Drop this into what you already run

TypeScript and Python SDKs, plus OpenTelemetry ingest — native for LangChain, Claude, Vercel AI, OpenClaw and LiveKit, with 15 framework integrations covered out of the box.

terminal
$ prefactor init

See it on your own agents

Book a demo and we'll walk through immutable audit trail on a fleet like yours — real frameworks, real traces.

Agent Performance Platform
Unified performance platform for agents, authentication, and risk management
All Systems Operational
3Global Agents
7Instances
5Services
12%Human Intervene
4High Risk
$2,360Monthly Spend
Mission ControlLive agent health with 7-day activity heartbeat
Claims Proc...68
$330/moRed
Claims Proc...65
$160/moRed
Claims Proc...82
$170/moAmber
ChatGPT74
$150/moAmber

See how every agent performs — and make it better

Prefactor helps teams observe, evaluate, and improve their AI agents in production — across every framework and provider.