Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.q-uestionable.ai/llms.txt

Use this file to discover all available pages before exploring further.

Use the qai audit subcommand to scan MCP servers, enumerate capabilities, and generate security reports. 
qai audit --help

​
qai audit scan

Scan an MCP server for security vulnerabilities across all 10 OWASP MCP Top 10 categories. 
qai audit scan [TARGET] [OPTIONS]
TARGET can be a URL (for SSE or Streamable HTTP servers) or a command string (for stdio servers). When provided, transport is inferred automatically and printed to the terminal. Use --transport to override the inference.

​
Options

FlagTypeRequiredDescription
TARGETpositionalNoServer URL or command. Prompted interactively if omitted.
--transportenumNoTransport type: stdio, sse, or streamable-http. Inferred from TARGET when omitted.
--commandstringNoServer command for stdio (alternative to positional TARGET)
--urlstringNoServer URL for SSE/HTTP (alternative to positional TARGET)
--checksstringNoComma-separated scanner names (e.g., injection,auth,token_exposure). Omit to run all 10.
--formatenumNoOutput format: json, sarif, html, ndjson, csv (default: json)
--outputstringNoOutput file path (default: results/scan.{format_ext})
--verbose / -vflagNoEnable debug logging

​
Transport Inference

When --transport is omitted, qai infers the transport from TARGET:
  • URL ending in /sse → SSE (high confidence, proceeds automatically)
  • URL without clear signal → prompts for transport choice in interactive terminals, fails with error in scripts
  • Command string (no ://) → stdio (high confidence, proceeds automatically)
The inferred transport is always printed: Inferred transport: sse (override with --transport).

​
Examples

Scan an SSE server (transport inferred from URL): 
qai audit scan http://localhost:3000/sse
Scan a stdio server: 
qai audit scan "npx @modelcontextprotocol/server-memory"
Explicit transport override: 
qai audit scan http://localhost:3000/mcp --transport streamable-http
Output as SARIF: 
qai audit scan http://localhost:3000/sse --format sarif --output audit-report.sarif
Run specific scanners only: 
qai audit scan "python my_server.py" --checks injection,auth,token_exposure
Backward-compatible flag syntax (still works): 
qai audit scan --transport stdio --command "python my_server.py"
Audit results are automatically saved to the database. Use the database ID returned in the output to retrieve or compare scan results over time.
Use --format sarif to generate reports compatible with GitHub Code Scanning integration.

​
qai audit enumerate

Enumerate MCP server capabilities without running security checks. Useful for discovery and asset inventory. 
qai audit enumerate [TARGET] [OPTIONS]
TARGET works the same as audit scan — URL or command string with automatic transport inference.

​
Options

FlagTypeRequiredDescription
TARGETpositionalNoServer URL or command. Prompted interactively if omitted.
--transportenumNoTransport type override: stdio, sse, or streamable-http
--commandstringNoServer command for stdio (alternative to positional TARGET)
--urlstringNoServer URL for SSE/HTTP (alternative to positional TARGET)

​
Examples

qai audit enumerate http://localhost:3000/sse
qai audit enumerate "npx @modelcontextprotocol/server-memory"
Output includes server name, protocol version, and lists of all tools, resources, and prompts.

​
qai audit list-checks

List all available scanner modules with their OWASP MCP Top 10 mappings. 
qai audit list-checks [OPTIONS]

​
Options

FlagTypeDescription
--frameworkstringShow framework IDs for a specific framework: owasp_mcp_top10, owasp_agentic_top10, mitre_atlas, cwe, or all

​
Example

qai audit list-checks --framework owasp_mcp_top10
Output is a table with scanner module name, category, description, status (Ready/Planned), and framework mappings if --framework is specified.

​
qai audit report

Generate a report from previously saved scan results in JSON format. 
qai audit report --input <path> [OPTIONS]

​
Options

FlagTypeRequiredDescription
--inputstringYesPath to saved scan results JSON file (from qai audit scan)
--formatenumNoOutput format: json, sarif, html, ndjson, csv (default: sarif)
--outputstringNoOutput file path (default: input path with new extension)

​
Example

qai audit report \
  --input results/scan.json \
  --format sarif \
  --output security-findings.sarif
Use qai audit report to regenerate findings in different formats without re-scanning the server. The --input file must be the JSON export from qai audit scan (not SARIF, HTML, or CSV). If you only have a non-JSON export, re-export the scan from the database first.