Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-31 · high
The TypeScript tool runs authorized pentests, bug bounty work and CTF labs locally, swapping OpenAI, Anthropic or DeepSeek as the model.
Why now: S1N6H posted the repo on August 26, and by August 31 it already wires in four model APIs plus auto-discovery for any OpenAI-compatible gateway.
Pentest Harness crossed 306 stars on GitHub after S1N6H posted it August 26. It's a TypeScript agent harness for authorized penetration tests, bug bounty work, security labs and CTFs. Sessions stay on your machine instead of routing through a vendor's cloud.
For researchers running those sessions, the model is a configuration line, not a vendor lock-in. The harness swaps between OpenAI's Chat Completions and Responses APIs, Anthropic's Messages API, DeepSeek, or any OpenAI-compatible gateway. It auto-discovers whatever models are on the other end.
The toolset covers what you'd want mid-engagement: shell access, filesystem operations, web research, a skills system, goal tracking, subagents, background jobs and workflow control. Subagents and background jobs let it run several steps of an engagement without someone babysitting each one.
The repo doesn't say how it performs against a real target, or how it guards against someone pointing it at systems outside authorized scope. For a tool that runs shell commands against live infrastructure, that gap matters before anyone relies on it.
Each link below shares sources, entities, or timing with this story.
OpenAI released Codex / Shared entities / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (OpenAI released Codex); both cover Anthropic Messages, OpenAI, Responses; reported by the same outlet (github.com).
OpenAI uses Claude Code / Shared entities / Same source domain / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (OpenAI uses Claude Code); both cover DeepSeek, TypeScript; reported by the same outlet (github.com).
OpenAI released Codex / Shared entities / Same source domain / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (OpenAI released Codex); both cover OpenAI, TypeScript; reported by the same outlet (github.com).
OpenAI supports MCP / Shared entities / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (OpenAI supports MCP); both cover DeepSeek, TypeScript; reported by the same outlet (github.com).
Linked by a graph relationship (OpenAI supports MCP); both cover Anthropic Messages, OpenAI Chat Completions; reported by the same outlet (github.com).
Linked by a graph relationship (OpenAI supports MCP); both cover DeepSeek, OpenAI; reported by the same outlet (github.com).
LLM uses OpenAI / Shared entities / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (LLM uses OpenAI); both cover DeepSeek, TypeScript; overlapping topics (agent, background, harness).
Anthropic partners with OpenAI / Shared entity: OpenAI / Same source domain / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Anthropic partners with OpenAI); both cover OpenAI; reported by the same outlet (github.com).