PARTNERS

Add selected Workbench capabilities through bounded OEM and partner integrations

Explore Partner Integrations
aisecurityllc
Sign inScope an AI Security Review
aisecurityllc
Scope an AI Security Review
← Deliverablesdeliverable
deliverable

RAG Test Harness Figures

Canonical figures for the RAG Test Harness.

Public sample
Client deliverable
public-sample
System
RAG Test Harness Figures
Environment
Production pilot

# RAG Test Harness Figures

RAG-01imported graph

Retrieval Trust Topology

RAG security depends on the full path from query and corpus boundaries through ranking, prompt assembly, model behavior, and output.

Graph showing a user query, retriever, tenant-filtered corpus, ranking, prompt assembly, model, optional tool, output, and provenance.

User queryUser and tenantidentityRetrieverTenant andaccess filterCorpusRanking andselectionPrompt assemblyModelOptional tool oractionOutputSourceprovenance
RAG-02coverage matrix

RAG Failure-Mode Coverage

A useful harness separates prompt injection, corpus poisoning, cross-tenant retrieval, provenance loss, and unsafe action propagation.

Matrix showing RAG failure modes across ingest, retrieval, prompt assembly, generation, action, and retest stages.

Indirect prompt injection

Hostile instructions enter through retrieved content.

Ingest
Covered
Retrieve
Covered
Assemble
Covered
Generate
Covered
Act
Partial
Retest
Covered

Corpus poisoning

Malicious or misleading content affects retrieval and generation.

Ingest
Covered
Retrieve
Covered
Assemble
Partial
Generate
Covered
Act
Partial
Retest
Covered

Cross-tenant retrieval

Content crosses an intended tenant or access boundary.

Ingest
Partial
Retrieve
Covered
Assemble
Covered
Generate
Covered
Act
Partial
Retest
Planned

Provenance loss

The system cannot preserve or communicate source origin and trust.

Ingest
Covered
Retrieve
Covered
Assemble
Covered
Generate
Partial
Act
Unknown
Retest
Covered

Unsafe action propagation

Retrieved content influences a tool or consequential action.

Ingest
Not applicable
Retrieve
Partial
Assemble
Covered
Generate
Covered
Act
Covered
Retest
Partial

Coverage states

CoveredPartialPlannedUnknownNot applicable
The matrix must be reviewed against the live harness. A named failure mode is not evidence that every pipeline stage is covered.
RAG-03

RAG Retest and Evidence Flow

A retrieval failure is closed only after corpus, ranking, policy, or action controls are changed and replayed.

RAG remediation lifecycle from failure reproduction through corpus or policy change, replay, evidence review, and closure or residual state.

RAG retest lifecycle
  1. 1
    Reproduce the failure
    Capture query, retrieved content, ranking, prompt assembly, model output, and action state.
  2. 2
    Identify the control boundary
    Determine whether the failure arises in ingest, retrieval, isolation, provenance, prompt assembly, or action policy.
  3. 3
    Change the control
    Update corpus, ranking, metadata, tenant policy, prompt policy, or tool policy.
  4. 4
    Replay the scenario
    Use the original and adversarial variants under the changed system.
  5. 5
    Review the evidence
    Confirm the intended boundary and inspect alternative failure paths.
Retest loop returns to Identify the control boundary
What did the retest prove?
  • Closed
  • Residual
  • Failed
  • Inconclusive