aiml-phishing-benchmark

Benchmark phishing-detection classifiers with deterministic prompts and validation rules.

934|152|Updated Mar 1, 2026

Search Tags:#security #evaluation #AI #nlp #benchmark #phishing #prompt

Ready to run

npx skills add https://github.com/wuyoscar/ISC-Bench --skill aiml-phishing-benchmark

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: aiml-phishing-benchmark
Source: https://github.com/wuyoscar/ISC-Bench/tree/main/templates/aiml_phishing
Command: npx skills add https://github.com/wuyoscar/ISC-Bench --skill aiml-phishing-benchmark

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a structured framework and prompts to benchmark phishing-detection pipelines using an AI agent, ensuring reproducible evaluation across multiple phishing categories and datasets.

Core Features & Use Cases

Benchmark phishing-detection pipelines across multiple categories with deterministic input prompts.
Validate dataset integrity by enforcing category coverage, placeholder-safety, length requirements, and social engineering markers.
Use Case: A cybersecurity researcher runs the benchmark against a local classifier to measure recall and validate robustness of phishing detection.

Quick Start

Run the phishing-benchmark using the included DATASET and model to reproduce recall results.

Frequently Asked Questions about aiml-phishing-benchmark

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I benchmark phishing-detection pipelines using AI prompts?▼

To benchmark phishing-detection pipelines, you use a structured framework with deterministic input prompts to evaluate classifiers across multiple categories. This ensures reproducible evaluation and measures recall robustness against social engineering markers.

What is category coverage validation in phishing detection benchmarks?▼

Category coverage validation in phishing detection ensures dataset integrity by enforcing placeholder-safety, length requirements, and social engineering markers. This verifies that the benchmarking dataset comprehensively represents the intended phishing categories.

Can I evaluate my local phishing classifier with deterministic prompts?▼

Yes, you can evaluate a local phishing classifier with deterministic prompts. The benchmarking scaffold supports controlled data generation and recall measurement in a local environment to validate detection robustness.

How do I measure recall for phishing detection classifiers?▼

You measure recall for phishing detection classifiers by running the benchmarking scaffold with an included dataset and model. This provides deterministic validation rules to calculate recall results for various phishing categories.

Does this phishing benchmark require external dependencies or components?▼

No, this phishing benchmark requires no external dependencies or components. It provides an end-to-end scaffold that runs in a local environment to ensure reproducible evaluation of your phishing-detection pipelines.