Skip to content

Add xihu-sword-ctf-agent: a deterministic-first CTF multi-agent framework #103

Description

@TrueFurina

Suggestion to add the following entry under the offensive/defensive agent tooling section:

- [xihu-sword-ctf-agent](https://github.com/TrueFurina/xihu-sword-ctf-agent) — A **deterministic-first** multi-agent CTF solving framework. A static-analysis layer (`presolve`, ~49 deterministic skills) runs first; the LLM is only a fallback behind a provider whitelist + token budget + wall-clock stop-loss. Ships an "honest water-level" discipline (KPI strictly separates `presolve` hits from `LLM` reasoning; no inflated numbers). MIT.

Why it fits this list:

  • Agent security governance: provider whitelist, token budget, and wall-clock stop-loss guardrails.
  • Auditable: every conclusion must be backed by a reproducible command + an on-disk artifact; narrative self-claims are rejected.
  • Honest water-level: it openly states "LLM autonomous solves on the real benchmark = 0" and documents its capability boundary instead of hiding it.

It's a concrete case study of governance practice for offensive-security agents, which I think readers of this list would value. Happy to open a PR adding the line directly if you'd like.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions