LLM and AI agent security resources.
Two documents, published openly at stable URLs. No form, no email gate. Forward them to whoever runs your security review.
Two documents. Both free.
Practical material on OWASP LLM01 prompt injection mitigation, written for engineers and security reviewers rather than for a landing page.
LLM Shield: Prompt Injection Detection Brief
What one call to the detection API returns: a risk score, the evidence spans that carry the attack, and a recommended action of PASS, LOG, FLAG or BLOCK. Covers why detection by persuasion mechanism survives paraphrasing, encoding and language swaps, the properties that hold on every request, and where the engine deploys.
ImposterHunter LLM Shield: Defense-in-Depth AI Security Platform Overview
The company, the problem, and the product at platform level: the surfaces LLM Shield covers as a defense-in-depth security layer, what the detection returns, the validation measured on the live API, and deployment options. Written to be forwarded internally to a security review.
Everything else lives on the product pages.
Rather than repackage the same material as a download, the detail is on the pages themselves, where it is indexable and linkable.
Five surfaces, one API.
Prompts, MCP tool responses, agent traffic, voice transcripts and retrieval pipelines. See LLM Shield.
Residency, subprocessors, compliance.
Data residency, the subprocessor list and the compliance roadmap are answered in full. Read the trust page.
Data Processing Addendum.
SCC-aligned, downloadable, counter-signed within five business days on request. Open the DPA.
Read the brief, then bring us your attacks.
A 30-minute call, live detection on real inputs, and an evaluation key so you can try to break it yourself.