reader_map.md

Reader map

Use this map to pick a fast path based on your role. Goal: get to a useful mental model in under 10 minutes.

If you are a security reviewer

Read in this order:

1. docs/threat_model_worked_example.md
2. docs/uvp_blackbox_interface.md
3. docs/THREAT_MODEL_BLACKBOX.md
4. docs/TEST_COVERAGE_MATRIX.md

What you should get:

• Threat model scope and assumptions.
• Blackbox control points (CreateClaim → Freeze → Execute → Capsule/ETL).
• Where leakage budgets, throttling, and freeze/escalation are tested.

If you are an ML eval engineer / researcher

Read in this order:

1. docs/EPISTEMIC_TRIAL_HARNESS.md
2. docs/uvp_blackbox_interface.md
3. examples/exfiltration_demo/
4. docs/threat_model_worked_example.md
5. docs/scenarios/

What you should get:

• A canonical entrypoint for running clinical-trial style epistemic evaluations.
• Expected blackbox I/O contract.
• How adaptive probing appears in practice and how controls respond.
• Scenario fixtures you can reuse for evaluation pipelines.

If you are in governance / policy

Read in this order:

1. docs/threat_model_worked_example.md
2. docs/uvp_blackbox_interface.md
3. docs/ALIGNMENT_SPILLOVER_POSITIONING.md
4. docs/TEST_EVIDENCE.md

What you should get:

• Plain-language explanation of why sequence-level controls matter.
• Decision/state vocabulary that can be mapped to policy gates.
• Evidence artifacts available for audits and incident review.

If you are a contributor

Read in this order:

1. README.md
2. docs/START_HERE.md
3. docs/README.md
4. docs/TEST_COVERAGE_PARAMETERS.md

Then run:

• cargo fmt --check
• cargo clippy --workspace --all-targets -- -D warnings
• cargo test --workspace

What you should get:

• Repo orientation, coding/testing expectations, and where to add new docs/examples safely.