Agentic attack‑tree generator on GitHub
An open GitHub repo published an AWS sample 'agentic attack tree' generator to help model and secure agentic AI workflows — a practical resource for teams building or defending AI agents. The repo surfaced alongside other AI security tooling discussions. (x.com)
The project appears under the aws-samples org as sample-agentic-attack-tree-generator and is presented on its site under the name “ThreatForest.” (aws-samples.github.io) ThreatForest orchestrates multiple autonomous AI agents using the Strands agentic framework to crawl a repository’s docs and code, extract context, and produce structured attack trees mapped to TTPs. (aws-samples.github.io) The tool uses a seven-stage pipeline (Setup → Discovery → Extraction → Generation → Enrichment → Mitigation → Reporting) and advertises typical run times of about 5–9 minutes depending on project size. (aws-samples.github.io) Generated outputs include interactive HTML dashboards (vis‑network), JSON exports, and mitigation recommendations that are enriched by automatic MITRE ATT&CK mapping. (aws-samples.github.io) The repo includes sample applications such as a cloud‑native microservices API and an HCLS example for testing the generator against realistic architectures. (github.com) Official docs and a Getting Started guide show pipx and pip installation options for quick local use, and the codebase is released under an MIT license. (aws-samples.github.io) Repository pages and recent commits show active development (a UI build file commit appeared roughly two weeks ago), and the project exposes a network/commit graph for community inspection. (github.com)
