Vision & Scope

Problem

Security researchers, penetration testers, and QA engineers need realistic HTTP traffic data for:

• Tool evaluation — testing how security scanners, WAFs, or API gateways handle diverse traffic
• Dataset generation — creating reproducible training/evaluation datasets for ML-based security tools
• Pipeline testing — exercising data pipelines (SIEM, log aggregators, threat detection) with varied HTTP data
• Demo environments — generating realistic traffic for product demos without real user data

Manually crafting hundreds of diverse HTTP requests is tedious, error-prone, and non-reproducible. Existing tools either focus on load testing (not diversity) or require complex scripting.

Solution

request-mock-skills provides a single, deterministic entry point for generating HTTP traffic data with:

• One command — generate everything from a single CLI invocation
• Deterministic output — same seed, same output, every time
• Rich diversity — 16 built-in transforms across 100+ HTTP dimensions
• Multiple formats — 13 export formats for any downstream tool
• Streaming delivery — handle large datasets without memory pressure
• Config-driven — YAML/JSON config files for repeatable batch generation

Status

Feature	Status	Description
Core generation	Done	Generate / GenerateOne converting RequestSpec to *har.Har
RequestSpec model	Done	100+ parameters across flat + 9 group structures
Templates	Done	20 built-in scenario templates
Transforms	Done	16 built-in transforms with pluggable signature
CLI	Done	11 subcommands with JSON envelope output
Export formats	Done	13 formats (curl, wget, python, postman, xml, yaml, har, json, jsonl, csv, markdown, html, text)
Batch generation	Done	Config-driven with deterministic seed
HAR post-processing	Done	split, export, dedup, add-headers, clone
Channel streaming	Done	GenerateStream / Engine.RunStream
URL decomposition	Done	composeURL for decomposed URL fields
Engine	Done	Engine unified entry point for Config
Request chain	Done	Chain for sequential multi-step requests
Documentation	In progress	VitePress documentation site

Future Directions

• More templates — additional protocol support (HTTP/2, HTTP/3 specific scenarios)
• More transforms — domain-specific transforms for security testing patterns
• Webhook delivery — stream generated entries to a webhook endpoint
• Plugin transforms — loadable shared library transforms
• OpenAPI/Swagger import — generate specs from API specifications
• HAR replay — replay generated HAR against a real target
• Performance benchmarks — benchmarks for large-scale generation (100k+ entries)
• Docker image — pre-built Docker images for CI/CD pipelines
• GitHub Actions — reusable action for CI integration

Design Principles

1. Determinism first — reproducibility is the primary design constraint
2. Progressive disclosure — simple CLI for common cases, full SDK for advanced use
3. Format agnostic — HAR is the intermediate format; export to anything
4. Composable — transforms, templates, and configs compose naturally
5. No external services — everything runs locally, no network required