FunversarialCV
Adversarial CV egg-injection console for LLM-driven hiring workflows.
See how a hidden instruction in a Word file can shift model behavior on the same CV — try the pre-injected demo, or start clean and configure payloads yourself.
Max 4 MB. DOCX (Word) only.
Contacts are replaced with placeholders before anything leaves your browser; nothing is stored.
FunversarialCV is an educational adversarial simulation for hands-on professional exploration, authorized security testing, and LLM research in hiring pipelines.
RUN THE CV EXPERIMENT
See what adversarial signals plant in a CV — and which ones a human never sees.
- Start with our sample CV (recommended)
or upload your own CV - Inject adversarial layers
- Download your "armed" CV
- Open the Validation Lab and run the ingestion lab on your baseline and armed builds—compare extractors, package metadata, and hyperlinks on this page for repeatable proof.
- Optional — to see how a live model reads the file, open the External comparative evaluation block. Public chat tools are an unreliable test (see the note in the Validation Lab).
- The repeatable result is in the Ingestion Lab, not an external chat — read what it surfaces for each signal.
- Decide what the file reveals about each signal.
This isn't about breaking the model — it's about understanding how inputs shape outcomes.
Signals are implemented as OWASP-aligned patterns; PII is dehydrated in your browser so outbound traffic carries tokens only; the service keeps zero retention after each response (authorized testing and research only).
Our server only sees tokenized placeholders (for example {{PII_EMAIL_0}}), not your raw email, phone, or address. Work is in-memory per request with no durable CV store; the file you save is rehydrated in your browser.
Full session · ~15–20 min — arm a CV and inspect every planted signal in the Ingestion Lab, baseline vs armed.