security-harness
Guide for the local-source VDH/VVS security research harness. Use when the operator wants authorized Web2/Web3 source review, MCP tools harness_plan harness_run harness_status, x-ray, solidity-auditor, fizz, or Pashov skills. Not a canned audit product and not a live scanner.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- security-harness
- license
- MIT
- description
- Guide for the local-source VDH/VVS security research harness. Use when the operator wants authorized Web2/Web3 source review, MCP tools harness_plan harness_run harness_status, x-ray, solidity-auditor, fizz, or Pashov skills. Not a canned audit product and not a live scanner.
- compatibility
- Requires a host that loads Agent Skills and MCP (Agent Plugins 1.0, or any CLI/cloud agent with those surfaces).
03 SKILL.md 正文
Security harness
This Skill teaches the host agent how to operate the harness. The host is whatever TUI or cloud agent loaded this plugin. The JSON worker is the backbone; it does not care which app spawned it. Worker models are bug-bounty hunters: Web2 or smart-contract bugs that an attacker could cash in. Impact first. MCP tools harness_models, harness_plan, harness_run, harness_status, and harness_evaluate spawn that worker. Do not invent a target, demo finding, or mock client.
First actions
- Confirm the operator has a real
scope.jsonbeside a tree they control. Required keys are in the harness README. There is no example target. - Call
harness_models. Pick three different catalog models (recon, hunter, validator). Hunter ≠ validator. Hunt compute is the{provider,id}pairs in theirscope.json. - Call
harness_planwith theirscope.json. Fix the scope until it succeeds. Web3 plans must list twelvehunt:web3:*cells. - Only then call
harness_runwithconfirmed=true, their scope, and an output directory outside the source root. That run is recon/x-ray → optional fizz → twelve auditor hunts → validate → VVS. - After a run:
harness_status. Confirmed findings stay 0 until a human reproduces.
x-ray, solidity-auditor, and fizz are names for harness stages. Call harness_run. Do not execute vendored Pashov playbooks (enumerate.sh, forge coverage, Echidna, Medusa). Models declare VDH/VVS tools; they cannot execute tools, patches, or tests.
04 同仓库更多 Skills
xitter
Interact with X/Twitter via the x-cli terminal client using official X API credentials. Use for posting, reading timelines, searching tweets, liking, retweeting, bookmarks, mentions, and user lookups.
writing-plans
Use when you have a spec or requirements for a multi-step task. Creates comprehensive implementation plans with bite-sized tasks, exact file paths, and complete code examples.
whisper
OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny (39M params) to large (1550M params). Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR.
weights-and-biases
Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform
webapp-testing
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
web-pentest
Authorized web pentest: recon, proof-based exploits, report.