Implement drift graph functionality in Explorer

[MuttleyD]

Context

Explorer is MIDAS’s graph-native surface for inspecting governance, authority, context, evidence, and runtime decision topology.

MIDAS already models decision surfaces, authority profiles, grants, agents, business services, capabilities, processes, evidence envelopes, and operational signals. It also has an existing direction around drift indicators and operational measures.

The next step is to make drift visible as a first-class graph capability in Explorer.

Problem

Agentic governance is not static. Decision behaviour, authority usage, evidence patterns, profiles, grants, and operating context can change over time.

Users need a way to see where drift is emerging, which decision surfaces are affected, which agents or profiles are involved, and whether the drift has operational or governance significance.

Without graph-level drift functionality, drift remains disconnected from the governance model that gives it meaning.

Scope

Design and implement an initial drift graph capability in Explorer.

The first version should focus on making existing drift-related data visible and useful, rather than inventing a new analytics platform.

The feature should explore how to represent drift across graph entities such as:

• Business Service
• Capability
• Process
• Decision Surface
• Authority Profile
• Grant
• Agent
• Evidence Envelope or runtime decision record

The implementation should identify the currently available drift data in the codebase and render it in a way that is consistent with the existing Context Graph and Authority Graph patterns.

Possible capabilities include:

• drift indicators on graph nodes;
• drift severity or status badges;
• drift-related graph filtering;
• a drift-focused graph view or overlay;
• inspector details explaining the drift signal;
• links from drift indicators to relevant evidence or runtime records;
• clear empty states when no drift data is available.

Acceptance Criteria

• Explorer includes an initial graph-based drift capability.
• The implementation is based on data already available in MIDAS or clearly documented as placeholder/demo-only where applicable.
• Drift is represented in relation to MIDAS governance entities, not as a disconnected list.
• Users can identify which decision surfaces, agents, profiles, or services are affected by drift.
• The visual treatment is consistent with existing Explorer graph patterns.
• Empty, loading, and error states are clear.
• The feature does not break existing Context Graph or Authority Graph behaviour.
• No new backend API is added unless necessary and justified.
• Any new API or data shape is documented.
• Existing relevant tests continue to pass.
• git diff --check passes.

Out of Scope

• Full drift analytics platform.
• Production-grade statistical drift detection.
• New machine-learning drift algorithms.
• Replacing the existing Context Graph or Authority Graph.
• Large-scale Explorer redesign.
• Enterprise compliance reporting.