National Cyber Warfare Foundation (NCWF)login

GreedyBear for extracting threat intelligence feeds from T-Pot honeypots


0 user ratings
2026-11-12 21:30:03
milo
Red Team (CNA)
"GreedyBear

GreedyBear is a defensive open-source platform that harvests attack data from T-Pot honeypot deployments and turns it into consumable threat intelligence feeds for detection and prevention.








ToolGreedyBear-Project/GreedyBear — threat intelligence platform that extracts attack data from T-Pot honeypots and generates IOC feeds
CategoryThreat intelligence / honeypot data enrichment (Python, MIT licensed)
Primary UseCollecting attacker IP indicators from authorized T-Pot honeypot deployments and publishing them as feeds such as recent.txt for defensive detection
Safe UsePurely defensive research tooling: intended for honeypot operators, SOC analysts, and authorized security teams running their own T-Pot infrastructure to block and detect scanning activity
Telemetry NoteGreedyBear sits on the defender's side — it ingests honeypot logs rather than touching production systems, and the public instance exposes feeds at greedybear.honeynet.org that defenders can monitor for their own ranges

GreedyBear is a defensive threat intelligence project hosted under the GreedyBear-Project organization on GitHub, written in Python and distributed under the MIT license. Its stated goal, taken directly from the README, is to extract attack data detected by a T-Pot honeypot — or an entire cluster of them — and to generate feeds that can be used to prevent and detect attacks. This positions it firmly in the blue-team corner of the security tooling landscape: rather than simulating or launching attacks, it harvests the telemetry that attackers voluntarily leave behind when they interact with decoy services.


For readers unfamiliar with the ecosystem, T-Pot is the well-known multi-honeypot distribution maintained by telekom-security, bundling honeypots such as Cowrie into a single sensor platform. GreedyBear acts as the aggregation and enrichment layer on top of it. The README even points to a live example feed drawn from Cowrie data, served at the public instance operated by The Honeynet Project at greedybear.honeynet.org, with a sample endpoint at /api/feeds/cowrie/all/recent.txt. That endpoint structure alone tells you the output format is plain-text IOC lists, the lingua franca of blocklists and SIEM watchlists.


The project's credibility is reinforced by its institutional backing. GreedyBear is a recognized project of The Honeynet Project, the long-running non-profit dedicated to open source security tooling and threat knowledge sharing, and the public feeds are provided for free through that sponsorship. The README also highlights participation in the 2026 Google Summer of Code, during which three projects were successfully implemented: an event collector API, a payload access feature, and dashboard modularization. That level of structured mentorship output is a strong signal of an actively governed project rather than a weekend experiment.


The origin story is documented with unusual honesty: it began as a personal Christmas project by Matteo Lodi in 2021, and grew through academic contribution. Tim Leonhard substantially improved the project and added machine learning models during his master thesis and is now the Principal Maintainer, while Martina Carella built the GUI as part of her own thesis work. For an operator evaluating whether to depend on this tool, that lineage matters — the ML enrichment and the web interface were not afterthoughts but deliberate research contributions that have since been folded into the maintained codebase.


From an engineering-hygiene perspective, the repository demonstrates modern Python practices. The badges show adoption of uv for dependency management, Ruff for linting, CodeQL for static security analysis, and automated dependency review and pull request automation workflows. For a tool whose entire purpose is to be trustworthy enough to feed into your defensive pipeline, visible CodeQL and dependency review gates are exactly what you want to see — supply-chain hygiene is non-negotiable for anything generating blocklists that firewalls will consume.


Architecturally, the README describes GreedyBear as a deployable service you run yourself and connect to your own T-Pot instance or cluster. The installation path runs through the project Wiki, and the documentation is organized around practical questions: how to try the public instance, how to dive into features, how to self-host, and how to stay current via the project blog with an RSS feed at feed.xml. This separation — a hosted public service for evaluation, and a self-hosted deployment for production sensor networks — mirrors the deployment model of mature open source intelligence platforms.


The feature set implied by the README spans three layers. First, ingestion: the event collector API, added during GSoC 2026, formalizes how attack events flow from honeypot sensors into the platform. Second, analysis: the machine learning models contributed during the thesis work presumably score and enrich the raw indicators, though the README does not enumerate the specific algorithms, so treat that as directionally accurate rather than a specification. Third, delivery: feeds segmented by honeypot type, as the cowrie feed path demonstrates, plus a dashboard GUI for exploration that was modularized in 2026.


Where does this fit in an authorized workflow? The canonical pattern is a research lab or enterprise SOC that operates one or more T-Pot sensors on IP space it owns, pointed at nothing production-critical. GreedyBear ingests the resulting attack sessions, deduplicates and aggregates the source addresses, and emits IOC feeds that can be pushed to perimeter defenses, threat hunting platforms, or SIEM correlation rules. Because every indicator originates from hostile interaction with a honeypot, false-positive risk on the indicators themselves is minimal — nobody legitimate is supposed to be attacking a decoy.


The public instance deserves its own note for defenders on a budget. The Honeynet Project's hosted deployment means you can consume community-generated feeds without running any infrastructure at all, which is a reasonable starting point for smaller teams. The tradeoff is standard for shared intelligence: you inherit the visibility of the sensors backing the public instance rather than the geographic and network specificity of your own decoys. Self-hosting against your own T-Pot cluster is the answer once feed fidelity to your threat landscape becomes a requirement.


Telemetry-wise, GreedyBear is a pure consumer of attacker behavior, not a generator of network activity against third parties. Defenders observing its operation would see normal API traffic between their honeypot sensors and the GreedyBear instance, and outbound consumption of recent.txt-style feeds by whoever subscribes. The payload access feature added in 2026 suggests captured malicious artifacts can also be retrieved through the platform for analysis, which is where standard malware-handling caution applies in your sandbox, not on your analyst workstation.


Community engagement appears healthy: the project carries a hacktoberfest topic, contribution guidelines live in the Wiki, and the maintainers actively route newcomers through issues and discussions before pull requests. With roughly 200 stars, an MIT license, and thesis-grade contributions, it sits in that sweet spot of established enough to trust, small enough that a motivated contributor can meaningfully shape it.


In summary, GreedyBear fills a real gap between raw honeypot logging and actionable intelligence. If you already run T-Pot, wiring it into GreedyBear is the natural next step to convert scattered attack sessions into structured, feed-consumable IOC output enriched with ML scoring and surfaced through a dashboard — all within a fully authorized, defensive research context. For teams without sensors, the public Honeynet Project feeds offer a zero-cost on-ramp to community honeypot intelligence.


For those wanting to evaluate it hands-on, two non-operational starting points from the README: browse the documentation in the GitHub Wiki under the Installation section, and inspect a live feed sample at the public instance's /api/feeds/cowrie/all/recent.txt endpoint to judge output quality before committing to a self-hosted deployment.



Official project repository for GreedyBear-Project/GreedyBear.

Download Tool

Educational analysis for authorized security professionals. Use only in controlled, authorized environments.






Source: OffensiveSec
Source Link: https://www.offsecblog.com/2026/11/greedybear-for-extracting-threat.html


Comments
new comment
Nobody has commented yet. Will you be the first?
 
Forum
Red Team (CNA)



Copyright 2012 through 2026 - National Cyber Warfare Foundation - All rights reserved worldwide.