creator/Anthropic-Cybersecurity-Skills
1
0
Fork 0
mirror of https://github.com/mukul975/Anthropic-Cybersecurity-Skills.git synced 2026-08-06 19:00:17 +03:00
Code Issues Packages Projects Releases Wiki Activity

Add 55 new skills across 3 new domains + 6 undercovered areas (762 -> 817)

Browse Source 
Demand-driven expansion targeting the fastest-growing 2025-2026 threat and
skills categories (ISC2/WEF/CrowdStrike/Mandiant signals):

- AI Security (NEW domain, 12 skills): LLM red-teaming with garak/PyRIT,
  prompt injection (direct/indirect/RAG), MCP tool-poisoning, agentic tool
  invocation, guardrails, model/data poisoning, system-prompt leakage,
  embedding/vector weaknesses, model extraction, continuous red-teaming
- Supply Chain Security (NEW domain, 5 skills): SBOMs, dependency confusion,
  malicious-npm triage, typosquatting, SLSA/Sigstore provenance
- Hardware & Firmware Security (NEW domain, 4 skills): CHIPSEC/UEFI audit,
  Secure Boot bypass, TPM measured-boot attestation, ESP bootkit hunting
- Identity (10): Entra ID/ROADtools, GraphRunner, AADInternals, ADCS/Certipy,
  shadow credentials, coercion, BloodHound CE, device-code phishing, SSO abuse
- Cloud-native (8): Stratus, Pacu, CloudFox, container escape, K8s RBAC,
  Falco, Trivy, kube-bench
- Offensive C2 (6): Sliver, Havoc, NetExec, DPAPI, NTLM relay ESC8, redirectors
- DFIR (6): Hayabusa, Chainsaw, KAPE, Velociraptor, EZ Tools, Plaso
- Backfill (4): OpenCTI, MISP, honeytokens, post-quantum crypto migration

Each skill follows the repo taxonomy (SKILL.md + references/{standards,api-reference}.md
+ scripts/agent.py + LICENSE), with researched real tool commands (no placeholders),
complete frontmatter, and ATT&CK/ATLAS + NIST CSF mappings. Updates README domain
table, skill count, and index.json.
This commit is contained in:
mukul975
2026-06-22 19:08:16 +02:00
parent 13a1c4afd9
commit 8cae0648ec
279 changed files with 36389 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files
skills/assessing-vector-and-embedding-weaknesses/LICENSE
+201
View File
@@ -0,0 +1,201 @@
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to the Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by the Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding any notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. Please do not remove or change
the license header comment from a contributed file except when
necessary.
Copyright 2026 mukul975
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
skills/assessing-vector-and-embedding-weaknesses/SKILL.md
+233
View File
@@ -0,0 +1,233 @@
---
name: assessing-vector-and-embedding-weaknesses
description: Test vector stores for embedding inversion, cross-tenant leakage, and poisoning.
domain: cybersecurity
subdomain: ai-security
tags:
- ai-security
- vector-database
- embedding-inversion
- rag-security
- owasp-llm08
- multi-tenant-isolation
- data-poisoning
- retrieval-augmented-generation
version: '1.0'
author: mahipal
license: Apache-2.0
nist_csf:
- MEASURE-2.7
mitre_attack:
- AML.T0024
---
# Assessing Vector and Embedding Weaknesses
> **Authorized use only:** These tests interact with vector stores and embedding models in RAG systems you own or are authorized to assess. Embedding inversion and cross-tenant probing against systems you do not control may expose third-party data and is prohibited without authorization.
## Overview
Retrieval-Augmented Generation (RAG) systems convert documents into embedding vectors stored in a vector database (Pinecone, Qdrant, Weaviate, Chroma, pgvector, FAISS) and retrieve the nearest vectors to ground LLM responses. OWASP **LLM08:2025 Vector and Embedding Weaknesses** covers the security risks unique to this layer:
- **Embedding inversion** — embeddings are not one-way. A trained inversion model (or a black-box reconstruction attack) can recover substantial portions of the original text from its vector, leaking source documents (maps to MITRE ATLAS **AML.T0024.001 Invert ML Model**).
- **Membership inference** — querying whether a specific record contributed to the corpus (AML.T0024.000).
- **Cross-tenant / multi-tenant leakage** — when one namespace/collection is shared or filter isolation is missing, a tenant retrieves another tenant's chunks.
- **Knowledge-base poisoning** — an attacker who can write to the corpus inserts crafted chunks that dominate retrieval (high cosine similarity to expected queries) and carry indirect prompt-injection payloads.
- **Retrieval manipulation** — adversarial documents tuned to be retrieved for many unrelated queries ("retrieval hijacking").
The parent technique is **AML.T0024 — Exfiltration via ML Inference API**: an attacker uses legitimate inference/query access to exfiltrate data (source text via inversion, membership, or model extraction). This skill provides a repeatable assessment of all five weakness classes.
## When to Use
- During a security assessment of any RAG / vector-search application (OWASP LLM08 coverage).
- When a vector store is multi-tenant and you must prove namespace/metadata isolation.
- When the corpus accepts user-supplied or third-party documents (poisoning surface).
- When the embedding endpoint is externally reachable (inversion/membership surface).
- When validating retrieval-filtering controls before go-live.
## Prerequisites
- Authorization and scope covering the target embedding endpoint and vector store.
- Python 3.10+.
- Read (and, for poisoning tests, write) access to a test collection — never the production corpus.
```bash
# Vector DB clients + embeddings + similarity tooling
python -m pip install numpy scikit-learn sentence-transformers
python -m pip install qdrant-client chromadb pinecone-client weaviate-client
# (optional) text-embedding inversion research baseline
python -m pip install vec2text
```
## Objectives
- Measure embedding-inversion exposure on the target embedding model.
- Run a membership-inference probe against the corpus.
- Test multi-tenant isolation (namespace, metadata filter, RBAC) for cross-tenant leakage.
- Inject benign poisoned chunks into a *test* collection and measure retrieval dominance.
- Detect indirect prompt-injection content surviving in retrieved chunks.
- Recommend controls: tenant-scoped filters, content validation, embedding-access limits.
## MITRE ATT&CK Mapping
| ID | Tactic | Official Technique Name | Role in this skill |
|----|--------|-------------------------|--------------------|
| AML.T0024 | ATLAS: Exfiltration | Exfiltration via ML Inference API | Using query/embedding access to exfiltrate source data |
| AML.T0024.000 | ATLAS: Exfiltration | Infer Training Data Membership | Membership-inference probe against the corpus |
| AML.T0024.001 | ATLAS: Exfiltration | Invert ML Model | Embedding-inversion reconstruction of source text |
| AML.T0020 | ATLAS: Resource Development | Poison Training Data | Knowledge-base poisoning of the corpus |
| AML.T0051.001 | ATLAS: Initial Access | LLM Prompt Injection: Indirect | Injection payloads embedded in retrieved chunks |
## Workflow
### Step 1: Inventory the RAG pipeline
Document the embedding model + dimensions, the vector store and its tenancy model, the chunking strategy, retrieval `top_k` and similarity metric (cosine/dot/L2), and any metadata filters applied at query time.
```python
# Example: inspect a Qdrant collection
from qdrant_client import QdrantClient
client = QdrantClient(url="http://localhost:6333")
info = client.get_collection("docs")
print(info.config.params.vectors) # size + distance metric
print(client.count("docs")) # corpus size
```
### Step 2: Test embedding-inversion exposure
Embeddings of similar text are close; an attacker with the embedding endpoint can iteratively reconstruct text whose embedding matches a target vector. Measure how much a nearest-neighbour-in-embedding-space recovers, using cosine similarity between candidate reconstructions and the target.
```python
import numpy as np
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
model = SentenceTransformer("all-MiniLM-L6-v2")
secret = "Patient John Doe, MRN 553120, diagnosed with hypertension."
target_vec = model.encode([secret])
# Attacker has only target_vec and the embedding endpoint. Hill-climb candidate text.
candidates = [
"Patient name and medical record number with a diagnosis.",
"John Doe medical record hypertension diagnosis",
"Patient John Doe MRN diagnosed hypertension",
]
cand_vecs = model.encode(candidates)
sims = cosine_similarity(target_vec, cand_vecs)[0]
for c, s in sorted(zip(candidates, sims), key=lambda x: -x[1]):
print(f"{s:.3f} {c}")
# High similarity for a near-verbatim guess => inversion risk is real for this model.
```
For a research-grade reconstruction baseline, `vec2text` can be used against compatible embedding models to demonstrate full-text recovery.
### Step 3: Membership inference
Determine whether a specific document is in the corpus by measuring the top-1 retrieval similarity for an exact-quote query: in-corpus items return a markedly higher max similarity than out-of-corpus controls.
```python
def membership_score(client, collection, embed, text):
vec = embed([text])[0].tolist()
hits = client.search(collection_name=collection, query_vector=vec, limit=1)
return hits[0].score if hits else 0.0
in_corpus = membership_score(client, "docs", model.encode, "<exact quote from a known chunk>")
control = membership_score(client, "docs", model.encode, "An unrelated random sentence.")
print(f"in-corpus={in_corpus:.3f} control={control:.3f} delta={in_corpus-control:.3f}")
# A large positive delta indicates the item is in the corpus (membership leak).
```
### Step 4: Test multi-tenant isolation
Confirm that tenant B cannot retrieve tenant A's chunks. Issue tenant-B-authenticated queries that *should* be filtered, and verify no tenant-A `tenant_id` appears in results.
```python
# Query as tenant B; expect ONLY tenant_id == "B" results.
from qdrant_client.models import Filter, FieldCondition, MatchValue
vec = model.encode(["confidential salary information"])[0].tolist()
hits = client.search(
collection_name="docs",
query_vector=vec,
limit=10,
query_filter=Filter(must=[FieldCondition(key="tenant_id", match=MatchValue(value="B"))]),
)
leaked = [h for h in hits if h.payload.get("tenant_id") != "B"]
print("CROSS-TENANT LEAK" if leaked else "isolation OK", "->", len(leaked), "foreign rows")
# Critical test: repeat WITHOUT the filter to confirm the server, not the client,
# enforces isolation. If unfiltered queries return tenant A data, isolation is client-side only.
hits_nofilter = client.search(collection_name="docs", query_vector=vec, limit=10)
print("server-side isolation FAILS" if any(h.payload.get("tenant_id") != "B" for h in hits_nofilter) else "OK")
```
### Step 5: Knowledge-base poisoning (test collection only)
Insert a benign poisoned chunk crafted to be retrieved for many unrelated queries, then measure how often it appears in `top_k`.
```python
from qdrant_client.models import PointStruct
# Benign marker payload (no real injection) to measure retrieval dominance.
poison = "POISON-CANARY. " + " ".join(
["password reset billing refund account login support error help"] * 8
)
client.upsert("docs_test", points=[
PointStruct(id=999999, vector=model.encode([poison])[0].tolist(),
payload={"tenant_id": "B", "source": "poison-test"})
])
queries = ["how do I get a refund", "reset my password", "what is the weather"]
for q in queries:
hits = client.search("docs_test", model.encode([q])[0].tolist(), limit=5)
dominated = any(h.payload.get("source") == "poison-test" for h in hits)
print(f"{'POISON in top5' if dominated else 'clean'}: {q}")
```
### Step 6: Detect indirect prompt injection in retrieved chunks
Scan retrieved chunk text for injection markers before it is concatenated into the prompt.
```python
import re
INJECTION_PATTERNS = [
r"ignore (all|previous|the above) instructions",
r"system prompt", r"you are now", r"disregard", r"</?(system|instructions)>",
]
def chunk_is_injection(text):
low = text.lower()
return [p for p in INJECTION_PATTERNS if re.search(p, low)]
for hit in client.search("docs", model.encode(["help"])[0].tolist(), limit=10):
flags = chunk_is_injection(hit.payload.get("text", ""))
if flags:
print("INDIRECT INJECTION in chunk", hit.id, flags)
```
### Step 7: Report and remediate
- **Inversion/membership:** rate-limit and authenticate the embedding endpoint; avoid returning raw similarity scores; restrict who can query embeddings.
- **Cross-tenant:** enforce tenant filters server-side (separate collections/namespaces per tenant where feasible); never rely on client-supplied filters.
- **Poisoning:** validate and provenance-tag every ingested chunk; scan inputs for injection; cap any single source's share of retrieval.
- **Indirect injection:** sanitize retrieved chunks and apply output guardrails (see `defending-llms-with-guardrails`).
## Tools and Resources
| Tool | Purpose | Primary Source |
|------|---------|----------------|
| OWASP LLM08 | Vector and Embedding Weaknesses guidance | https://genai.owasp.org/llmrisk/llm082025-vector-and-embedding-weaknesses/ |
| sentence-transformers | Embedding generation for testing | https://www.sbert.net/ |
| Qdrant client | Vector store + filtered search | https://qdrant.tech/documentation/ |
| Chroma / Weaviate / Pinecone | Alternative vector stores | https://docs.trychroma.com/ |
| vec2text | Embedding-inversion research baseline | https://github.com/jxmorris12/vec2text |
| MITRE ATLAS | AML.T0024 Exfiltration via ML Inference API | https://atlas.mitre.org/ |
## Validation Criteria
- [ ] RAG pipeline inventoried (embedding model, store, tenancy, metric, top_k, filters).
- [ ] Embedding-inversion exposure measured and rated.
- [ ] Membership-inference delta computed for in-corpus vs control items.
- [ ] Multi-tenant isolation tested both with and without client filters (server-side enforcement confirmed).
- [ ] Poisoning dominance measured in a test collection only.
- [ ] Retrieved chunks scanned for indirect-injection content.
- [ ] Findings reported with remediation for each weakness class.
- [ ] No production corpus modified during the assessment.
skills/assessing-vector-and-embedding-weaknesses/references/api-reference.md
+51
View File
@@ -0,0 +1,51 @@
# API and Command Reference
## sentence-transformers (embedding generation)
| Call | Purpose |
|------|---------|
| `SentenceTransformer("all-MiniLM-L6-v2")` | Load an embedding model (384-dim) |
| `model.encode([texts])` | Return numpy array of embeddings |
| `model.encode(text, normalize_embeddings=True)` | L2-normalized vectors (for cosine) |
## scikit-learn similarity
| Call | Purpose |
|------|---------|
| `cosine_similarity(a, b)` | Pairwise cosine similarity matrix |
## Qdrant client (qdrant-client)
| Call | Purpose |
|------|---------|
| `QdrantClient(url="http://localhost:6333")` | Connect |
| `client.get_collection(name)` | Inspect vector size + distance metric |
| `client.count(name)` | Corpus size |
| `client.search(collection_name, query_vector, limit, query_filter)` | k-NN search with optional filter |
| `client.upsert(name, points=[PointStruct(id, vector, payload)])` | Insert/update points |
| `Filter(must=[FieldCondition(key, match=MatchValue(value))])` | Metadata filter (tenant isolation) |
## Chroma (chromadb)
| Call | Purpose |
|------|---------|
| `chromadb.Client()` / `PersistentClient(path)` | Connect |
| `collection.query(query_embeddings=[...], n_results=k, where={...})` | k-NN with metadata filter |
| `collection.add(ids, embeddings, metadatas, documents)` | Insert |
## Pinecone (pinecone-client)
| Call | Purpose |
|------|---------|
| `Pinecone(api_key=...)` | Connect |
| `index.query(vector=..., top_k=k, namespace="tenant", filter={...})` | k-NN; namespace = tenant boundary |
| `index.upsert(vectors=[(id, vec, meta)], namespace=...)` | Insert |
## Assessment metrics
| Metric | Meaning |
|--------|---------|
| Inversion cosine | Similarity between reconstructed candidate and target vector; high = recoverable. |
| Membership delta | top-1 score(in-corpus query) − top-1 score(control query); large positive = membership leak. |
| Poison dominance | Fraction of unrelated queries returning the poison chunk in top_k. |
| Cross-tenant count | Number of foreign-tenant rows returned to a tenant query (should be 0). |
## vec2text (research baseline)
| Call | Purpose |
|------|---------|
| `vec2text.load_pretrained_corrector("gtr-base")` | Load inversion corrector for compatible embedder |
| `vec2text.invert_embeddings(embeddings, corrector)` | Reconstruct text from embeddings |
skills/assessing-vector-and-embedding-weaknesses/references/standards.md
+35
View File
@@ -0,0 +1,35 @@
# Standards and Framework Mapping
## NIST AI Risk Management Framework (AI RMF 1.0 / GenAI Profile NIST AI 600-1)
| ID | Name | Rationale |
|----|------|-----------|
| MEASURE-2.7 | AI system security and resilience are evaluated and documented | Assessing inversion, membership, isolation, and poisoning weaknesses measures the security/resilience of the RAG vector layer. |
## MITRE ATLAS
| ID | Name | Rationale |
|----|------|-----------|
| AML.T0024 | Exfiltration via ML Inference API | Parent technique: query/embedding access is abused to exfiltrate source data. |
| AML.T0024.000 | Infer Training Data Membership | Membership-inference probe determines whether a record is in the corpus. |
| AML.T0024.001 | Invert ML Model | Embedding inversion reconstructs source text from vectors. |
| AML.T0020 | Poison Training Data | Knowledge-base poisoning inserts adversarial chunks into the corpus. |
| AML.T0051.001 | LLM Prompt Injection: Indirect | Injection payloads surviving in retrieved chunks. |
## OWASP Top 10 for LLM Applications (2025)
| ID | Name | Rationale |
|----|------|-----------|
| LLM08 | Vector and Embedding Weaknesses | The core risk class under test (inversion, leakage, poisoning). |
| LLM02 | Sensitive Information Disclosure | Inversion/membership leakage discloses sensitive source data. |
| LLM01 | Prompt Injection | Indirect injection delivered through poisoned retrieval. |
## Weakness class to control mapping
| Weakness | Control |
|----------|---------|
| Embedding inversion | Authenticate + rate-limit embedding endpoint; avoid exposing raw scores. |
| Membership inference | Restrict similarity-score exposure; add query auditing. |
| Cross-tenant leakage | Server-side tenant filters or per-tenant collections/namespaces. |
| Knowledge-base poisoning | Provenance tagging, content validation, per-source retrieval caps. |
| Indirect injection in chunks | Sanitize retrieved text; apply output guardrails. |
skills/assessing-vector-and-embedding-weaknesses/scripts/agent.py
+152
View File
@@ -0,0 +1,152 @@
#!/usr/bin/env python3
"""Vector/embedding weakness assessor.
Runs four checks against a RAG pipeline you are authorized to test:
inversion - measures how close a guessed reconstruction sits to a target vector
membership - computes in-corpus vs control top-1 similarity delta (Qdrant)
isolation - verifies server-side tenant filtering (Qdrant)
injection - scans retrieved chunk text for indirect prompt-injection markers
Examples
--------
python agent.py inversion --secret "MRN 553120 hypertension" \
--guess "patient MRN hypertension diagnosis"
python agent.py membership --url http://localhost:6333 --collection docs \
--quote "<exact chunk quote>" --control "unrelated sentence"
python agent.py isolation --url http://localhost:6333 --collection docs \
--tenant-field tenant_id --tenant B --query "salary information"
"""
import argparse
import re
import sys
INJECTION_PATTERNS = [
r"ignore (all|previous|the above) instructions",
r"system prompt", r"you are now", r"disregard previous",
r"</?(system|instructions)>",
]
def get_embedder(model_name):
try:
from sentence_transformers import SentenceTransformer
except ImportError:
sys.exit("[!] sentence-transformers not installed. pip install sentence-transformers")
return SentenceTransformer(model_name)
def get_qdrant(url):
try:
from qdrant_client import QdrantClient
except ImportError:
sys.exit("[!] qdrant-client not installed. pip install qdrant-client")
return QdrantClient(url=url)
def cmd_inversion(args):
from sklearn.metrics.pairwise import cosine_similarity
model = get_embedder(args.model)
tv = model.encode([args.secret])
gv = model.encode([args.guess])
sim = float(cosine_similarity(tv, gv)[0][0])
print(f"inversion cosine(secret, guess) = {sim:.4f}")
verdict = "HIGH inversion risk" if sim >= 0.85 else "moderate" if sim >= 0.6 else "low"
print(f"verdict: {verdict}")
return 1 if sim >= 0.85 else 0
def cmd_membership(args):
model = get_embedder(args.model)
client = get_qdrant(args.url)
def top1(text):
vec = model.encode([text])[0].tolist()
hits = client.search(collection_name=args.collection, query_vector=vec, limit=1)
return hits[0].score if hits else 0.0
in_corpus = top1(args.quote)
control = top1(args.control)
delta = in_corpus - control
print(f"in-corpus top1 = {in_corpus:.4f}")
print(f"control top1 = {control:.4f}")
print(f"delta = {delta:.4f}")
leak = delta >= 0.2
print("verdict:", "MEMBERSHIP LEAK" if leak else "no clear membership signal")
return 1 if leak else 0
def cmd_isolation(args):
from qdrant_client.models import Filter, FieldCondition, MatchValue
model = get_embedder(args.model)
client = get_qdrant(args.url)
vec = model.encode([args.query])[0].tolist()
filtered = client.search(
collection_name=args.collection, query_vector=vec, limit=10,
query_filter=Filter(must=[FieldCondition(
key=args.tenant_field, match=MatchValue(value=args.tenant))]),
)
leak_filtered = [h for h in filtered
if h.payload.get(args.tenant_field) != args.tenant]
unfiltered = client.search(collection_name=args.collection, query_vector=vec, limit=10)
leak_unfiltered = [h for h in unfiltered
if h.payload.get(args.tenant_field) != args.tenant]
print(f"filtered query foreign rows : {len(leak_filtered)}")
print(f"unfiltered query foreign rows : {len(leak_unfiltered)}")
if leak_filtered:
print("CRITICAL: filtered query returned other tenants' data")
elif leak_unfiltered:
print("WARNING: isolation appears client-side only; server returns cross-tenant data")
else:
print("isolation OK at server level for this query")
return 1 if (leak_filtered or leak_unfiltered) else 0
def cmd_injection(args):
model = get_embedder(args.model)
client = get_qdrant(args.url)
vec = model.encode([args.query])[0].tolist()
hits = client.search(collection_name=args.collection, query_vector=vec, limit=args.limit)
found = 0
for h in hits:
text = (h.payload or {}).get("text", "")
flags = [p for p in INJECTION_PATTERNS if re.search(p, text.lower())]
if flags:
found += 1
print(f"INJECTION in chunk {h.id}: {flags}")
print(f"{found} of {len(hits)} retrieved chunks flagged for indirect injection")
return 1 if found else 0
def main():
p = argparse.ArgumentParser(description="Vector/embedding weakness assessor")
p.add_argument("--model", default="all-MiniLM-L6-v2", help="embedding model")
sub = p.add_subparsers(dest="cmd", required=True)
s = sub.add_parser("inversion"); s.add_argument("--secret", required=True); s.add_argument("--guess", required=True)
s = sub.add_parser("membership")
s.add_argument("--url", required=True); s.add_argument("--collection", required=True)
s.add_argument("--quote", required=True); s.add_argument("--control", required=True)
s = sub.add_parser("isolation")
s.add_argument("--url", required=True); s.add_argument("--collection", required=True)
s.add_argument("--tenant-field", default="tenant_id"); s.add_argument("--tenant", required=True)
s.add_argument("--query", required=True)
s = sub.add_parser("injection")
s.add_argument("--url", required=True); s.add_argument("--collection", required=True)
s.add_argument("--query", default="help"); s.add_argument("--limit", type=int, default=10)
args = p.parse_args()
dispatch = {
"inversion": cmd_inversion, "membership": cmd_membership,
"isolation": cmd_isolation, "injection": cmd_injection,
}
try:
sys.exit(dispatch[args.cmd](args))
except Exception as exc:
sys.exit(f"[!] {args.cmd} failed: {exc}")
if __name__ == "__main__":
main()