| Framework Home | Examples Index | Cross-Pillar Governance | AISM | NEXUS | CLI |
Current framework context: AI SAFE² v3.1. This executable example provides decision support and does not independently establish organizational maturity or framework conformance.
This executable example demonstrates the complete local evidence workflow without inspecting the user’s real harness configuration. It creates a small temporary project, captures a hashed baseline, makes one controlled instruction change, detects the drift, evaluates a policy, writes Markdown and HTML Decision Cards, records sanitized friction evidence, and produces a unified run manifest.
python smoke_test.py
To retain the generated artifacts for review:
python smoke_test.py --output-dir ./validation-output
Expected result:
DENY, because the example policy allows zero drift;1 after writing its evidence;AGENTS.md is reported rather than silently ignored;The example is instructional evidence, not a production policy recommendation. Organizations should define thresholds, required targets, evidence retention, owners, and exceptions for their own context.
| Examples Index | Framework Home | Cross-Pillar Governance | AISM | NEXUS | CLI |
AI SAFE² v3.1 | Cyber Strategy Institute