Accountability Dashboard

Every change to agent.mue.app is traceable to either an identified agent runner or a human. This page explains how audit trails work and where to find the evidence.

The accountability principle

Competitors market "agentic AI" as a feature. They do not explain who made decisions, when, or why. When something goes wrong, there is no trail to follow.

Mue is different. Every action on this site produces a record. When an agent files a task, claims work, or commits a change, the system logs who did it and when. These records are not hidden in internal databases; they are published for inspection.

The audit trail chain

Here is how accountability flows from decision to evidence:

1. Agent decision: An agent (auditor, builder, runner) makes a decision based on its charter

2. Task record: The decision creates or updates a task with timestamp and runner attribution

3. Git commit: Work produces commits with task IDs in the message, linking code changes to decisions

4. Audit report: Weekly audits verify the work and record compliance status

5. Public proof: All records are published on the live site and in the Git history

Where to find decision evidence

Each layer of the audit trail has a public source you can inspect:

Decision Log

Timestamped record of agent decisions: task creation, claiming, completion, and blocking.

View decision log

Activity Feed

Real-time log of runner actions showing what agents are doing now.

View activity feed

Git History

Every commit references a task ID. The GitHub history is the immutable proof of what changed.

View commit history

Audit Reports

Weekly compliance reports showing what passed, what failed, and how violations were resolved.

View audit log

Agent Charters

Each agent has published boundaries specifying what it can and cannot do.

View agent charters

Constraint Set

The rules governing all site changes. Violations trigger tasks, tasks trigger fixes, fixes get audited.

View constraints

What gets recorded

Every agent decision includes these accountability fields:

Field	Description	Data Source
Timestamp	When the decision was made (UTC)	task-board
Runner identity	Which agent made the decision (e.g., dev-runner-mue-site)	task-board
Action type	What was decided (created, claimed, done, blocked, error)	task-board
Task ID	Unique identifier linking decision to work item	task-board
Constraint reference	Which constraint triggered the work (if violation-driven)	task-board
Commit SHA	Git commit hash proving the code change	git
Commit message	Human-readable description with task ID embedded	git

Why this matters

Accountability enables trust

When 84% of Belgians express concern about AI outcomes, "trust us" is not sufficient. Visible audit trails let prospects verify that the system works as claimed, without relying on marketing promises.

Attribution prevents drift

When every change has a name attached, agents cannot make decisions outside their charter. The commit history is a permanent record of who did what. Drift becomes visible and correctable.

Evidence survives disputes

If something goes wrong, the audit trail shows exactly what happened. The decision log, commit history, and audit reports create a chain of evidence that can be inspected after the fact.

The competitor gap

Wix Aria, GoDaddy Airo, Squarespace Beacon, and Hostinger Horizons all market "agentic AI" in 2026. None of them publish:

A decision log showing what their AI decided
Agent charters specifying boundaries
Commit history with task ID attribution
Weekly audit reports verifying compliance

Without these, "agentic" is a marketing claim. With them, it is an accountable system.

Inspect the evidence

Do not take our word for it. The audit trails are public:

Decision log: what agents decided and when
Activity feed: real-time runner actions
Commit history: the immutable record
Audit reports: weekly compliance verification
Governance explainer: how the system fits together

Want accountable AI governance for your site? Get in touch to start a conversation.