Add tenant/document ACL tests to the evaluation suite
Jump & context
Week 5: RAG ACLs, PII UUID Gateway, Training-Data Poisoning & Agent/MCP Foundations
The five hours
- 01
Concepts & system design
60 min· read · map the mechanismR03 — OWASP API authorization
R13 — Qdrant payload filtering & security
Read the assigned sections; map the mechanism, threat model and trade-offs.
- 02
Project implementation
120 min· build · integrate · debugIntegrate tenant/document ACL assertions and poisoned-document checks directly into the automated 30-pair RAG evaluation runner.
Build, integrate, debug and improve the increment in the active project.
- 03
Security lab
60 min· attack your own systemRun the evaluation suite under multiple tenant identities with unauthorized and poisoned chunks present in the store.
Test the abuse case, failure path, authorization bypass or resource limit.
- 04
Verification & evidence
30 min· test · record evidenceCombined quality + security regression report for
P2retrieval.Add or run regression tests; record the 9-point evidence record.
- 05
Deliberate practice & Socratic review
30 min· explain aloud · update gapsAnswer the 8 Socratic questions without notes: What? Why? How? What if? Why not? How can I prove it? Where can it break? What have I learned?
Re-explain the hardest concept aloud, inspect one edge case, update the gap log.
- total5 h
Read today · inside the concepts block
R03 R13BOLA (API1), Broken Auth (API2), Unrestricted Resource Consumption (API4), SSRF (API7), Misconfiguration (API8).
Collections, payload metadata indexing, pre-filtering, multitenancy partitioning, and snapshots.
Read only the named sections needed for today's task. Reading sits inside the 60-minute concepts block; it is not extra time.
Free resources for this day
5 curated- docsQdrant Documentation · reference docsMultitenancy
Explains payload-based partitioning with per-tenant indexes versus collection-per-tenant, the isolation model your ACL assertions must verify on every retrieval call.
R13checked 2026-11-11 - articleMedium (Haydar Külekci) · technical articleMulti-Tenant Vector Search in Practice
Shows the tenant filter running inside HNSW traversal rather than after it, which is the property that makes a cross-tenant result a filter bug rather than a ranking accident.
R13R03unchecked 2026-11-11 - videoThe Cyber Expert · 17 minAPI1 - Broken Object Level Authorization | IDOR | crAPI
Exploits BOLA against crAPI by swapping object identifiers, the same substitution attack your automated ACL test pairs should perform against document_id values.
R03checked 2026-11-11 - videofreeCodeCamp.org · 87 minOWASP API Security Top 10 Course – Secure Your Web Apps
Covers each API risk including authorization and resource consumption, giving the assertion categories to add to the evaluation runner beyond functional quality.
R03checked 2026-11-11 - labOWASP (GitHub) · self-hosted labOWASP/crAPI: completely ridiculous API
A runnable intentionally vulnerable API with BOLA challenges, so you can practice writing cross-tenant assertions that fail loudly before porting them into the 30-pair runner.
R03checked 2026-11-11
Evidence record · 9-point standard
A lab is complete only when it produces evidence. Installing PyRIT, Garak, Presidio or TransformerLens is not completion. Saved on this device only.
Day notes
emptyThe two source formulations
verbatim · nothing summarisedProject & Stage Target
P2 | S4
Core Implementation
Integrate tenant/document ACL assertions and poisoned-document checks directly into the automated 30-pair RAG evaluation runner.
Security Lab
Run the evaluation suite under multiple tenant identities with unauthorized and poisoned chunks present in the store.
Verification & Evidence
Combined quality + security regression report for P2 retrieval.
Read Today (60m Concepts Block)
R03 (OWASP API authorization), R13 (Qdrant payload filtering & security).
---