# manyruumpus/privacy-aware-retrieval-for-secure-rag-systems- architecture

> Architecture diagram of the GitHub repository manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-18). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-
- Repository: https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-

## Overview

SecureRAG-Bench is a research benchmark for comparing baseline vector RAG, access-controlled retrieval, and heuristic query sanitization on synthetic data. A caller identity and question enter the guarded pipeline; retrieval selects documents, context is sent to local Ollama generation, and results are scored for privacy and utility. Separate dataset builders and development/held-out experiment runners support the benchmark. The runtime and selected scripts are evidenced by excerpts and the source index; other documented capabilities are represented at their existing script paths without assuming their internal wiring.

Main entry points: [`src/secure_rag/guarded_pipeline.py`](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/guarded_pipeline.py) (Guarded pipeline), [`src/secure_rag/pipeline.py`](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/pipeline.py) (RAG pipeline), [`scripts/run_heldout_v2.py`](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_heldout_v2.py) (Held-out runner), [`src/secure_rag/retrieval.py`](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/retrieval.py) (Vector retrieval)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Benchmark data"]
  node_corpus_build["Corpus builder<br/>[build_corpus.py]"]
  node_query_build["Query builder"]
  node_heldout_build["Holdout builder<br/>[build_heldout.py]"]
end

subgraph group_runtime["RAG runtime"]
  node_guard["Query guard<br/>[query_guard.py]"]
  node_guarded["Guarded pipeline"]
  node_pipeline["RAG pipeline<br/>[pipeline.py]"]
  node_retrieval["Vector retrieval<br/>[retrieval.py]"]
end

subgraph group_evaluation["Experiments and evaluation"]
  node_cli["Interactive CLI<br/>[chat.py]"]
  node_dev_runner["Development runner<br/>[run_experiment.py]"]
  node_v2_runner["V2 runner"]
  node_heldout_runner["Held-out runner<br/>[run_heldout_v2.py]"]
  node_answer_audit["Answer audit"]
  node_metrics["Metrics<br/>[compute_metrics.py]"]
end

subgraph group_integrations["External models"]
  node_embedding{{"Sentence transformer"}}
  node_ollama{{"Ollama"}}
end

node_caller(("Caller"))

node_caller -->|"submits query"| node_guarded
node_guarded -->|"sanitizes query"| node_guard
node_guarded -->|"delegates run"| node_pipeline
node_pipeline -->|"searches documents"| node_retrieval
node_retrieval -->|"encodes text"| node_embedding
node_pipeline -->|"requests generation"| node_ollama
node_corpus_build -.->|"supplies corpus"| node_query_build
node_heldout_build -->|"reuses builder"| node_query_build
node_cli -.->|"runs queries"| node_guarded
node_dev_runner -.->|"runs experiments"| node_guarded
node_v2_runner -.->|"runs variants"| node_guarded
node_heldout_runner -->|"runs holdout"| node_guarded
node_heldout_runner -->|"uses prompt pilot"| node_v2_runner
node_answer_audit -.->|"audits scores"| node_metrics

click node_corpus_build "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_corpus.py"
click node_query_build "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_evaluation_queries.py"
click node_heldout_build "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_heldout.py"
click node_guard "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/query_guard.py"
click node_guarded "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/guarded_pipeline.py"
click node_pipeline "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/pipeline.py"
click node_retrieval "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/retrieval.py"
click node_cli "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/chat.py"
click node_dev_runner "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_experiment.py"
click node_v2_runner "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_experiment_v2.py"
click node_heldout_runner "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_heldout_v2.py"
click node_answer_audit "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/audit_answer_scoring.py"
click node_metrics "https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/compute_metrics.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_corpus_build,node_query_build,node_heldout_build toneBlue
class node_guard,node_guarded,node_pipeline,node_retrieval toneAmber
class node_cli,node_dev_runner,node_v2_runner,node_heldout_runner,node_answer_audit,node_metrics toneMint
class node_embedding,node_ollama toneRose
class node_caller toneIndigo
```

## Components

### Benchmark data

- [Corpus builder](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_corpus.py): `scripts/build_corpus.py`
- [Query builder](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_evaluation_queries.py): `scripts/build_evaluation_queries.py`
- [Holdout builder](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/build_heldout.py): `scripts/build_heldout.py`

### RAG runtime

- [Query guard](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/query_guard.py): `src/secure_rag/query_guard.py`
- [Guarded pipeline](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/guarded_pipeline.py): `src/secure_rag/guarded_pipeline.py`
- [RAG pipeline](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/pipeline.py): `src/secure_rag/pipeline.py`
- [Vector retrieval](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/src/secure_rag/retrieval.py): `src/secure_rag/retrieval.py`

### Experiments and evaluation

- [Interactive CLI](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/chat.py): `scripts/chat.py`
- [Development runner](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_experiment.py): `scripts/run_experiment.py`
- [V2 runner](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_experiment_v2.py): `scripts/run_experiment_v2.py`
- [Held-out runner](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/run_heldout_v2.py): `scripts/run_heldout_v2.py`
- [Answer audit](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/audit_answer_scoring.py): `scripts/audit_answer_scoring.py`
- [Metrics](https://github.com/manyruumpus/privacy-aware-retrieval-for-secure-rag-systems-/blob/main/scripts/compute_metrics.py): `scripts/compute_metrics.py`

### External models

- **Sentence transformer**
- **Ollama**

### Other components

- **Caller**

## Connections

- Caller → Guarded pipeline: submits query (evidence: `README.md`)
- Guarded pipeline → Query guard: sanitizes query (evidence: `src/secure_rag/guarded_pipeline.py`)
- Guarded pipeline → RAG pipeline: delegates run (evidence: `src/secure_rag/guarded_pipeline.py`)
- RAG pipeline → Vector retrieval: searches documents (evidence: `src/secure_rag/pipeline.py`)
- Vector retrieval → Sentence transformer: encodes text (evidence: `src/secure_rag/retrieval.py`)
- RAG pipeline → Ollama: requests generation (evidence: `src/secure_rag/pipeline.py`)
- Corpus builder → Query builder: supplies corpus
- Holdout builder → Query builder: reuses builder (evidence: `scripts/build_heldout.py`)
- Interactive CLI → Guarded pipeline: runs queries
- Development runner → Guarded pipeline: runs experiments
- V2 runner → Guarded pipeline: runs variants
- Held-out runner → Guarded pipeline: runs holdout (evidence: `scripts/run_heldout_v2.py`)
- Held-out runner → V2 runner: uses prompt pilot (evidence: `scripts/run_heldout_v2.py`)
- Answer audit → Metrics: audits scores
