PhD prototype showcase

CAPRA

A Context-Aware Privacy Risk Assessment framework, implemented as an executable prototype, that connects agent telemetry, contextual reasoning, bounded risk inferences, proposed refinement, and human review.

Review Gamma View the repository

Gamma · completed research iteration

Survey-informed refinement

Gamma is the completed, bounded research iteration of CAPRA. In the expert survey, 41 valid respondents assessed the framework: all 15 criterion items met the predefined consensus rule, with agreement from 82.9% to 97.6%. Feedback drove refinements to terminology, interfaces, context governance, human oversight, change governance, telemetry controls, and evidence visibility.

The survey provides design feedback and consensus evidence. It is not described as framework validation or operational-effectiveness evidence.

What Gamma implements

Six conceptual layers remain: DFL, CPL, CL, RIL, FRL, and HIL. Five executable stages—Federate, Contextualise, Assess, Refine, and Review—operate over the shared Context Layer. External Systems remain outside CAPRA.

Governed Context Records retain provenance, purpose, candidate obligations, uncertainty, ownership, and validation state. Candidate obligations guide human review; they are not compliance findings.

Review Gamma source and evidence →
Human reviewEarly context review supports Confirm, Correct, Reject, or Defer; post-assessment review records a durable decision linked to upstream evidence.
Change controlVersioned schemas, mappings, criteria, models, prompts, and policies are selected through governed artefact records, with role separation, status history, application, and rollback.
TelemetryAllowlisting, redaction, role-gated detail, synthetic public views, configurable retention, legal hold, and deletion receipts demonstrate minimisation, access, and retention mechanics.
EvidenceA deterministic synthetic admissions walkthrough retains source-to-review lineage and checksum-verifiable outputs. Twenty-five controlled positive and negative tests exercise the bounded profile.
Claim boundary: Gamma does not establish semantic correctness, legal compliance or applicability, privacy effectiveness, operational reliability, or production readiness. Publication of the research artefact and this explanatory page does not make Gamma a deployed service or a production-ready software release.

Five-part showcase

1

Research project outline

The research question, gap, and contributions in seven slides.

2

Framework description

The six layers of CAPRA, one at a time, culminating in a sped-up end-to-end animation.

3

Conceptual CAPRA demonstration

A visual walkthrough of how CAPRA connects federation, contextualisation, risk assessment, adaptation, and human oversight.

4

CAPRA prototype walkthrough

A practical walkthrough of the six-layer privacy risk assessment prototype for agentic AI systems.

5

Prototype repository

The repository includes the bounded Gamma implementation and its deterministic admissions evidence, plus historical Alpha/Beta workflows, synthetic domain fixtures, local infrastructure, dashboards, and retained test evidence. The Docker reviewer route remains Beta-era historical implementation evidence.

Open on GitHub → Read the quick start

Example instantiation · university admissions

Gamma's authoritative walkthrough uses a deterministic synthetic admissions event and retains a complete identifier-linked evidence chain through early context review, assessment, refinement, post-assessment review, application, and rollback. The historical Beta reviewer route separately uses twelve synthetic applicant records. The admissions fixtures are distinct from the retained environmental_air_quality.csv fixture used for an ad hoc Beta stress test.

CAPRA reference architecture lifecycle showing Federate, Contextualise, Assess, and Adapt connected by data, context, assessment, and refinement flows, with human oversight informing adaptation
CAPRA reference architecture lifecycle. The admissions prototype instantiates this lifecycle as the five executable stages below over the shared Context Layer.
FederateNormalise generated events into telemetry_raw records with canonical event identifiers.
ContextualiseAdd actor, purpose, data, and transmission context, with immediate provenance links.
AssessConstruct scenarios and store bounded risk inferences for reviewer inspection.
RefinePrepare proposed changes with an action, target, rationale, and expected effect.
ReviewPresent an Approve or Reject form that resumes the waiting n8n execution.
Gamma provides the deterministic complete admissions evidence chain described above. Separately, a historical Beta local-fallback window produced records through refinement but did not submit the review form. That Beta run remains stage-level historical evidence and is not a Gamma result.

The framework at a glance

Select a layer to inspect its Gamma responsibility and the retained Beta implementation evidence. The layers are conceptual responsibilities; the five executable stages run over the shared Context Layer.

Data Federation Layer

Treats External Systems as outside CAPRA, validates their boundary records, and minimises admitted telemetry through allowlisting, redaction, and quarantine.

ReceivesUntrusted generated events and source-system telemetry.
ContributesMinimised canonical records with stable event identifiers.
Historical Beta evidencetelemetry_raw records in MongoDB.

Architecture in motion

The animation illustrates CAPRA's designed processing path from an external source through federation, contextualisation, assessment, refinement, and human review. It shows the implemented workflow structure rather than an empirically verified same-event trace across every store.

Animated illustration of CAPRA's designed processing path through its six architectural layers
The animation retains the implementation labels used when the reference workflow was captured.

Domains evaluated

Historical campaigns exercised higher-education, healthcare, and retail inputs through the same CAPRA architecture. Domain-classified records were confirmed through risk assessment, with shared refinement output in aligned observation windows. The beta route provides new deterministic synthetic fixtures for these domains.

Healthcare
Clinicians, records, treatment vs. curiosity — special-category health data.
Retail
Customer profiles, loyalty accounts, third-party sharing risk.
Student admission
Applications, transcripts, personal statements, reviewer workflows.

Run the historical Beta reviewer route

Start the no-build reviewer route

Docker provisions n8n, MongoDB, Fuseki, Loki, and Grafana. Supply one generic OpenAI-compatible endpoint, or select the local llama3.2 fallback.

git clone https://github.com/gracebilliris/capra-prototype.git
cd capra-prototype
cp reviewer/env.template reviewer/.env

# Add your endpoint values to reviewer/.env, then:
./reviewer/scripts/bootstrap.sh
./reviewer/scripts/verify_route.sh
./reviewer/scripts/run_demo.sh --domain admissions --minutes 30

For the credential-free fallback, set CAPRA_LLM_PROVIDER=ollama and follow the host or container instructions in the quick start.

Open complete instructions →

What opens locally

n8nlocalhost:5679 · workflow and executions
Grafanalocalhost:3002 · privacy-risk dashboard
Fusekilocalhost:3031 · shared Context Layer
Lokilocalhost:3101 · stage logs queried through Grafana
The endpoint route is provider-neutral. Credentials stay in a git-ignored local environment file and are imported into the local n8n credential store. No browser-based credential wiring is required.

Related publications

Published peer-reviewed and preprint outputs that motivate, ground, or extend the CAPRA prototype.

Featured publication · A-ranked journal

A Federated Observability Architecture Pattern for Reliable Agentic AI Software Systems Across the AI Software Development Lifecycle

Published in Information and Software Technology, Volume 199, Article 108260 (2026). The paper presents a seven-layer architecture pattern for federating and observing heterogeneous telemetry across the AI software development lifecycle.

Read the paper View the prototype

Published

  1. Billiris, G., & Gill, A. (2026). A Federated Observability Architecture Pattern for Reliable Agentic AI Software Systems Across the AI Software Development Lifecycle. Information and Software Technology, 199, Article 108260. https://doi.org/10.1016/j.infsof.2026.108260
  2. Billiris, G., Gill, A., Haggag, O., Bandara, M., & Grundy, J. (2026). CPL: A Context Processing Layer for Semantic Observability in Multi-Agent AI Systems. SSRN 7194446. https://doi.org/10.2139/ssrn.7194446
  3. Billiris, G., Gill, A., & Bandara, M. (2026). Systematic Literature Review of Data Privacy Risks in AI Systems. Science and Information Computing Conference (SAI) 2026. https://link.springer.com/book/10.1007/978-3-032-24810-7
  4. Billiris, G., Gill, A., & Bandara, M. (2025). Privacy in the Age of AI: A Taxonomy of Data Risks. arXiv. https://doi.org/10.48550/arXiv.2510.02357
  5. Billiris, G., Gill, A., & Bandara, M. (2025). A Taxonomy of Data Risks in AI and Quantum Computing (QAI): A Systematic Review. arXiv. https://doi.org/10.48550/arXiv.2509.20418
  6. Billiris, G., Gill, A., Oppermann, I., & Niazi, M. (2024). Towards the Development of a Copyright Risk Checker Tool for Generative Artificial Intelligence Systems. Digital Government: Research and Practice, 5(4), Article 41. https://doi.org/10.1145/3703459
  7. Billiris, G., & Gill, A. Q. (2024). An Initial Review of the Copyright Concerns of Generative Artificial Intelligence. ACIS 2024 Proceedings, Article 17. https://aisel.aisnet.org/acis2024/17/