Probes vector database endpoints to detect unauthenticated exposure of embeddings and RAG data.
Also known as GuardBee/guardbee-mcp, @guardbee/mcp-vector-store-scanner
- Stars
- 2
- Tools
- 2
- Downloads, 30 days
- 1,887
- Latest version
- 0.1.1
- Last commit
- Nov 12, 2026
- License
- MIT
What is vector-store-scanner MCP?
vector-store-scanner probes vector database endpoints to detect unauthenticated exposure of embeddings and RAG data. It exposes 2 tools, such as scan_endpoint, list_patterns.
Summary generated from the project's README
About vector-store-scanner MCP server
Security professionals and developers can use this tool to identify unauthenticated exposure of embeddings and RAG data within vector database endpoints. By probing network targets, it detects if sensitive information is reachable without authentication, mimicking the path an attacker would take to find misconfigured instances. It works through an MCP client like Claude Desktop or via a CLI for CI/CD integration.
The scanner supports six target types: Weaviate, Qdrant, Chroma, Elasticsearch/OpenSearch, Redis, and Postgres/pgvector. It uses escalating checks that move from retrieving instance information to listing schemas and finally attempting to read actual stored data. For Redis and Postgres, it utilizes native wire protocols like RESP and the Postgres wire protocol to verify authentication requirements without ever sending or guessing credentials.
Users can configure the scanner using a guardbee.yml file to set timeout limits or define which severity levels should trigger a failure. The tool provides output in SARIF 2.1.0 format for integration into automated scanning workflows. It focuses specifically on unauthenticated read access rather than testing the permissions of valid API keys.
What you can do with it
- 1Detect unauthenticated vector data exposureCheck if a specific Qdrant or Weaviate endpoint allows anyone to read stored documents without a password.
- 2Audit database schema visibilityVerify if your data models and collection names are visible to unauthorized users via HTTP probes.
- 3Automate security scanning in CI/CDRun a command line scan on internal endpoints and output the results in SARIF format for GitHub Code Scanning.
- 4Verify Redis and Postgres securityEnsure that Redis and Postgres instances require authentication by testing their native wire protocols.
Written from the project's README. Every use case is checked against a quote from it.
How to install vector-store-scanner MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall vector-store-scanner MCP in Claude Code
$ claude mcp add vector-store-scanner -- npx -y @guardbee/mcp-vector-store-scannerInstall vector-store-scanner MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"vector-store-scanner": {
"command": "npx",
"args": [
"-y",
"@guardbee/mcp-vector-store-scanner"
]
}
}
}Install vector-store-scanner MCP in VS Code
.vscode/mcp.json
{
"servers": {
"vector-store-scanner": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@guardbee/mcp-vector-store-scanner"
]
}
}
}Install vector-store-scanner MCP in VS Code CLI
$ code --add-mcp '{"name":"vector-store-scanner","type":"stdio","command":"npx","args":["-y","@guardbee/mcp-vector-store-scanner"]}'Install vector-store-scanner MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"vector-store-scanner": {
"command": "npx",
"args": [
"-y",
"@guardbee/mcp-vector-store-scanner"
]
}
}
}Install vector-store-scanner MCP in Codex
$ codex mcp add vector-store-scanner -- npx -y @guardbee/mcp-vector-store-scannerInstall vector-store-scanner MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.vector-store-scanner] command = "npx" args = ["-y", "@guardbee/mcp-vector-store-scanner"]
Install vector-store-scanner MCP in Gemini CLI
$ gemini mcp add -s user vector-store-scanner npx -- -y @guardbee/mcp-vector-store-scannerInstall vector-store-scanner MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"vector-store-scanner": {
"command": "npx",
"args": [
"-y",
"@guardbee/mcp-vector-store-scanner"
]
}
}
}Started by HostedMCP on 16 Nov 2026
We ran this command for @guardbee/mcp-vector-store-scanner 0.1.1 in an isolated container with no network and no keys. It answered and listed 2 tools.
vector-store-scanner MCP tools 2
Read from the server with tools/listscan_endpointProbe a vector-database (or vector-index-capable) endpoint for unauthenticated exposure β Weaviate, Qdrant, Chroma, Elasticsearch/OpenSearch (HTTP fingerprinting, escalating from instance info β collection/schema listing β actual stored data), Redis (RESP PING without AUTH), and Postgres/pgvector (wire-protocol SSLRequest + StartupMessage handshake, no real credentials sent). Read-only β never writes or authenticates with real credentials.
list_patternsList the vector-store types this scanner can fingerprint, and what unauthenticated-access checks it runs for each
Automated checks
Ran Nov 13, 2026Full report βPassed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablePass
- Open source license (OSI)Pass
- Commit or release in the last yearPass
- Package name not close to a popular packagePass
- Remote endpoint uses HTTPSNot checked
- Destructive tools are labelledPass
- No hidden instructions to the modelPass
- No critical advisory in the package or its dependenciesPass
README
View on GitHub@guardbee/mcp-vector-store-scanner
π¬π§ English | πΉπ· TΓΌrkΓ§e | π¨π³ δΈζ | π―π΅ ζ₯ζ¬θͺ | π°π· νκ΅μ΄ | π§π· PortuguΓͺs | πͺπΈ EspaΓ±ol | π©πͺ Deutsch
An MCP (Model Context Protocol) server that probes a vector-database (or vector-index-capable) endpoint for unauthenticated exposure β the same class of misconfiguration behind repeated real-world incidents of open Elasticsearch/MongoDB/Redis instances leaking data, now aimed at the RAG-era stack: Weaviate, Qdrant, Chroma, Elasticsearch/OpenSearch, Redis, and Postgres/pgvector.
Every RAG pipeline's embeddings live somewhere. If that vector store is reachable on the network without authentication, anyone who finds it can read (and often write) every document your retrieval pipeline was ever given β including anything sensitive that made it into a chunk.
Telemetry is off by default. With
GUARDBEE_TELEMETRY=1it sends the server, tool, success, duration and short parameter values such as a file path; longer strings and fields likecontent,tokenorpasswordare replaced with a placeholder. See@guardbee/mcp-telemetry.
Claude βββΊ vector-store-scanner βββΊ Your vector-store endpoint
β
ββ weaviate / qdrant / chroma / elasticsearch (HTTP fingerprint)
β info βββΊ schema/collections βββΊ actual stored data
ββ redis (raw RESP: PING, no AUTH sent)
ββ postgres (raw wire protocol: SSLRequest + StartupMessage, no password sent)
How it escalates
Each HTTP-based probe doesn't stop at "is this thing up" β it walks the same path an attacker would, in increasing order of impact, and only reports what actually succeeded:
- Instance info (medium) β version/build details readable without auth. Low impact alone, but confirms the service is exposed and helps fingerprint known CVEs.
- Schema / collection listing (high) β class or collection names readable without auth. Reveals your data model even before any real data is read.
- Actual stored data (critical) β a real object/point/document comes back without auth. This is the finding that matters: your embeddings (and whatever they were built from) are readable by anyone who can reach this port.
Redis and Postgres get the same treatment via their native wire protocols instead of HTTP β a PING (Redis) or a StartupMessage (Postgres) is enough to learn whether authentication is enforced, without ever sending a real password. No credential is ever guessed, brute-forced, or sent β every check here is read-only fingerprinting, the same category of probing dns-intelligence and ssl-inspector already do elsewhere in this monorepo.
Features
- 6 target types: Weaviate, Qdrant, Chroma, Elasticsearch/OpenSearch (HTTP), Redis (raw RESP), Postgres/pgvector (raw wire protocol)
- Auto-detection β fingerprints the endpoint by trying each known type in turn when
typeisn't specified - From-scratch protocol implementations for Redis and Postgres β no client library dependency, no real credentials ever sent, includes a TLS-upgrade path for Postgres's
SSLRequesthandshake - Escalating checks (info β schema β data) so a "critical" finding always means real data was actually read, not just that a port responded
- SARIF 2.1.0 output β CI/CD integration (GitHub Code Scanning, etc.)
- Config file support via
guardbee.yml - 22 unit tests against local mock HTTP/TCP servers, covering every probe's matched/unmatched fingerprint and each escalation level
Quick Start
Claude Desktop / MCP Client
{
"mcpServers": {
"guardbee-vector-store-scanner": {
"command": "npx",
"args": ["-y", "@guardbee/mcp-vector-store-scanner"]
}
}
}
CLI (CI/CD)
npx @guardbee/mcp-vector-store-scanner scan my-qdrant.internal --type=qdrant --fail-on=high --format=sarif > results.sarif
MCP Tools
| Tool | Description |
|---|---|
scan_endpoint |
Probes a single host/port for unauthenticated exposure |
list_patterns |
Lists the supported store types and what each check looks for |
What it catches
| Store | Check | Severity |
|---|---|---|
| Weaviate | /v1/meta readable without auth |
medium |
| Weaviate | /v1/schema readable without auth β reveals class names |
high |
| Weaviate | /v1/objects returns real data without auth |
critical |
| Qdrant | / instance info readable without auth |
medium |
| Qdrant | /collections readable without auth |
high |
| Qdrant | /collections/<name>/points/scroll returns real data without auth |
critical |
| Chroma | heartbeat reachable without auth | medium |
| Chroma | /collections readable without auth |
high |
| Elasticsearch/OpenSearch | cluster info readable without auth | medium |
| Elasticsearch/OpenSearch | /_cat/indices readable without auth |
high |
| Elasticsearch/OpenSearch | /<index>/_search returns real documents without auth |
critical |
| Redis | PING accepted without AUTH |
critical |
| Postgres/pgvector | no TLS offered | medium |
| Postgres/pgvector | AuthenticationOk with no password challenge |
critical |
Configuration (guardbee.yml)
vector-store-scanner:
fail-on: high # any | critical | high | medium | low | none
timeout-ms: 4000
Limitations (by design)
- Chroma's actual-document read isn't checked β its data-read API varies significantly across versions, so this scanner stops at the collection-listing level for it (still enough to detect the primary "no auth at all" misconfiguration).
- Milvus isn't supported β its primary API is gRPC, which doesn't fingerprint via a lightweight probe the way the others do.
- This checks for unauthenticated read access, not authorization bugs behind a working auth layer (e.g. a valid-but-overprivileged API key) β that's out of scope for an external, credential-less probe.
Development
npm run build
npm test # 22 unit tests
License
MIT β GuardBee
Feedback
Questions, ideas, or a false positive? Start a thread in GitHub Discussions or open an issue. If GuardBee is useful to you, a β on the repository helps others find it.
vector-store-scanner MCP questions
How do I add vector-store-scanner MCP to Claude Code?
Run claude mcp add vector-store-scanner -- npx -y @guardbee/mcp-vector-store-scanner in your terminal. Add --scope user to make it available in every project.
How do I add vector-store-scanner MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does vector-store-scanner MCP provide?
It exposes 2 tools, including scan_endpoint, list_patterns.
Does the scanner attempt to guess or brute-force passwords?
No, the scanner performs read-only fingerprinting and does not guess, brute-force, or send real credentials.
Is telemetry enabled by default?
No, telemetry is turned off by default unless you explicitly enable it using the GUARDBEE_TELEMETRY=1 environment variable.
What kind of security vulnerabilities does it detect?
It detects unauthenticated read access where sensitive embeddings or data models are exposed without authentication.