otparse-mcp
A containerized MCP server for parsing OT/ICS packet captures, specifically supporting Modbus and BACnet protocols via tshark. It provides structured JSON output from PCAP files to enable LLMs to analyze industrial network traffic and derive device inventories.
README
otparse
A containerized MCP server that parses OT/ICS packet captures and returns structured JSON. Designed to plug into a Claude Code workflow for analyzing saved PCAP files from Modbus/TCP and BACnet/IP networks.
Status: [EXPERIMENTAL]
What It Does
otparse exposes two MCP tools:
parse_modbus_pcap-- reads a saved PCAP file, extracts Modbus/TCP transactions using tshark, and returns decoded frames plus a basic device inventory derived from observed IPs.parse_bacnet_pcap-- reads a saved PCAP file, extracts BACnet/IP packets using tshark, and returns decoded frames plus a basic device inventory derived from observed IPs.
Both tools accept a path to a PCAP inside the container and an optional packet limit. Output is structured JSON that an LLM can reason over directly.
Background
This grew out of interest in OT/ICS security and the gap between "I have a PCAP from an industrial network" and "I can actually understand what devices were talking to each other and whether anything looks off." Wrapping tshark in an MCP server means Claude can do the first layer of that analysis without a separate toolchain.
The container design follows the same pattern as netparse: read-only evidence mount, no outbound network access, non-root user.
Requirements
- Docker (for the containerized path)
- Or: Python 3.11+, tshark installed on the host (for local dev)
Setup
Container (recommended)
docker build -t otparse-mcp .
Local dev
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
pip install -e .
tshark must be available in your PATH for local development.
Usage
Run as a containerized MCP server
docker run --rm -i \
--network none \
-v /srv/evidence:/evidence:ro \
otparse-mcp:latest
Claude Desktop / Claude Code config
{
"mcpServers": {
"otparse": {
"command": "docker",
"args": [
"run", "--rm", "-i",
"--network", "none",
"-v", "/srv/evidence:/evidence:ro",
"otparse-mcp:latest"
]
}
}
}
Example tool call
Once connected, ask Claude something like:
"Parse the Modbus capture at
/evidence/modbus/example.pcapand summarize what devices were communicating."
Claude will call parse_modbus_pcap and work from the returned JSON.
Run locally (dev only)
python -m otparse.server
Project Structure
otparse/
├── otparse/
│ ├── __init__.py
│ ├── server.py # FastMCP server, tool definitions
│ ├── models.py # Pydantic models for all inputs/outputs
│ ├── parsers/
│ │ ├── common.py # Shared helpers (path validation, tshark check, type coercion)
│ │ ├── modbus.py # Modbus/TCP parser using pyshark + tshark
│ │ └── bacnet.py # BACnet/IP parser using pyshark + tshark
│ └── analyzers/
│ └── devices.py # Builds device inventory from parsed transactions
├── Dockerfile
├── docker-compose.yml
├── pyproject.toml
├── requirements.txt
└── README.md
Tool Reference
See docs/tool-reference.md for full parameter and output documentation.
Notes / Limitations
- This tool only analyzes saved PCAP files. It does not do live capture.
- tshark does the actual protocol dissection. If a PCAP uses a non-standard port or an unusual encapsulation, tshark may not identify the traffic and the tool will return an empty result with a note.
- Modbus
direction(request vs. response) detection depends on pyshark exposing the right attributes. Some captures may produce"unknown"for most frames. - The device inventory is derived purely from observed source/destination IPs. It does not correlate with vendor OUIs, DHCP records, or any external data source.
- BACnet device instance numbers and object names are extracted when present, but tshark attribute names vary by version. Some fields may be
nullon older tshark builds. docker-compose.ymlis included for convenience but assumes the repo root layout matches the build context. Adjustbuild:paths if your layout differs.- Both tools accept an optional
packet_limitparameter (default: 500) that caps the number of matching packets decoded per run. This limit is in place during early testing and will be removed in a future version. For large captures, increase the limit explicitly when calling the tool.
Test PCAPs
Sample ICS PCAPs suitable for testing can be found in the ITI ICS-Security-Tools repository, which maintains a collection of ICS protocol captures for security research.
Planned
extract_ics_devicestool combining Modbus and BACnet inventory into a single view- Anomaly heuristics (unusual function codes, broadcast storms, write-heavy sessions)
- Timeline/session summarizer tool
- Fixture PCAPs and parser unit tests
Changelog
See CHANGELOG.md
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.