violin - Supervised, Hermes-native penetration testing prof

项目说明

<p align="center">
<img src="assets/logo.png" alt="" width="256"/>
</p>

<h1 align="center"> ☤ — Supervised Agentic Hermes Pentest Profile</h1>

<p align="center">
<a href=" src=" alt="Release Ready"></a>
<a href=" src=" alt="License: MIT"></a>
<a href=" src=" alt="Hermes >=0.18.0"></a>
<a href=" src=" alt="Kali Linux"></a>
<a href=" src=" alt="Parrot OS"></a>
<a href=" src=" alt="Landing page"></a>
<a href=" src=" alt="GitHub stars"></a>
</p>

<p align="center">
<b>35 playbooks · 19 references · 14 templates · required execution guard · Hermes-native</b>
</p>

is a Hermes-nati pentest profile for supervised, authorised penetration tests — from reconnaissance through safe exploit to reporting. It uses Hermes' built-in toolsets, se skills, and the required plugin at the target-execution boundary. The standalone CLI supports release checks, diagnostics, and administrati; target commands run through the plugin. adds no profile-specific credentials and inherits the provider and tool backends already configured in Hermes.

<p align="center">
<a href="#quick-start">Quick start</a> ·
<a href="#engagement-lifecycle">Workflow</a> ·
<a href="#guard-tools">Guard tools</a> ·
<a href="#benchmarks">Benchmarks</a> ·
<a href=" page</a> ·
<a href=" ·
<a href="#de">De</a>
</p>

<p align="center">
<strong>Following ?</strong>
<a href=" the repository</a>
to keep it easy to find and help other security engineers disco agentic pentesting.
</p>

Community discovery

is listed in community-curated collections including
awesome-ai-security,
awesome-hermes-skills,
awesome-hermes-agent, and
awesome-infosec.

Third-party co:
Starlog technical analysis and
“: Supervised Agentic Penetration Testing Profile for Hermes Agent”.

For independent e, use the
re kit, which
collects the benchmark path, known limits, review context, and a factual
comparison with autonomous pentest-agent designs.

---

<table>
<tr><td width="240"><strong>Guarded target execution</strong></td><td>Scope, phase, PTT, skill, hypothesis, history, and synchronization checks run before a target command starts.</td></tr>
<tr><td><strong>Persistent engagement state</strong></td><td>PTT tasks, hypotheses, command history, checkpoints, e, and reports survi compression.</td></tr>
<tr><td><strong>E findings</strong></td><td>A typed submission binds each finding to authenticated execution receipts.</td></tr>
<tr><td><strong>Routed methodology</strong></td><td>A pentest orchestrator selects focused web, identity, API, business-logic, LLM-security, and misconfiguration playbooks.</td></tr>
<tr><td><strong>Bounded execution</strong></td><td>Single commands, command bursts, background processes, batch review, heartbeat checks, and cancellation share one state model.</td></tr>
<tr><td><strong> releases</strong></td><td>Plugin registration, schemas, skill snapshots, documentation contracts, lint, formatting, and the full test suite are release-gated.</td></tr>
</table>

Quick start

Install the profile

Then start with an authorized target and let collect the scope before
any target interaction:

Requirements

does not select a model or provider. Configure those in Hermes. For a
capable default, use Qwen3.8 27B locally or DeepSeek V4 Flash through a
hosted provider.

Engagement lifecycle

1. Initialize the engagement and approve scope/scope.yaml.
2. Select one active PTT task and its routed skill with
.
3. Run target commands with or .
4. Update hypotheses as e changes their status.
5. Review each bounded command batch with .
6. Submit findings and generate the final report.
7. Complete the retrospective.

The complete phase model is:

Starting work in a new phase requires a PTT task under that phase. Existing
tasks are not mo phase sections.

Guard tools

The plugin registers twel tools from one typed registry:

| Tool | Purpose |
|---|---|
| | Create, start, refresh, close, or cancel a PTT task |
| | Create or update a scoped hypothesis |
| | Submit a finding bound to its signed execution receipts |
| | Execute one guarded command |
| | Execute a bounded command file |
| | Read background execution status |
| | Cancel tracked background execution |
| | Review a completed batch and settle state |
| tch | Rebind a pending batch after confirmation |
| | Clear a completed heartbeat review |
| | Resolve the appro target |
| | Explain current tasks, skills, and blockers |

is the generic target-command boundary. There are no
tool-specific execution adapters or binary allowlists. Installed
non-interacti may run only after the engagement gates pass.

The raw-terminal hook is a best-effort safety net, not network containment.
Use terminal only for host-local preparation and administration.

Safety model

The detailed policy is
skills/pentest/references/standards.md.

Engagement state

init-engagement creates the canonical working structure:

Skill delivery

Skills are loaded on demand. The first call for a routed
skill may return skill_prepared without changing the PTT. After Hermes
delivers the skill content, repeat the transition to bind the receipt and apply
the task change. reports the exact recovery action.

Marker files such as .skill-loaded-* do not pro delivery.

Routed skills

| Skill | Co |
|---|---|
| pentest | Scope, lifecycle, recon, exploitation, reporting, closeout |
| web-app | Injection, SSRF, tra, deserialization, client-side flaws |
| identity-auth | Authentication, authorization, IDOR, JWT, CSRF, cryptography |
| api-testing | REST, SOAP, GraphQL, WebSocket |
| business-logic | Workflow, pricing, coupon, quota, referral, race conditions |
| llm-security | Prompt injection, MCP, JSON-RPC |
| misconfig | Deployment, error handling, obser, obscurity |

The orchestrator loads only the phase material and specialist playbook needed
for the current task.

Administrative CLI

The CLI does not replace guarded target execution.

check-command exposes admission checks for diagnostics and does not execute
the supplied command.

Architecture

Benchmarks

The repository includes the Escape Duck Store definition plus known-good and
known-bad scorer fixtures. Calibration proves that the scorer handles those
fixtures; it does not establish li recall, workflow completion, or
report quality.

Read the benchmark methodology before publishing a score.
It defines the proof, reproducibility, private e, and disclosure
requirements for a credible result.

De

The release gate checks surfaces, isolated plugin import, registered
tools, generated schemas, skill snapshots, documentation contracts, Ruff, and
the full test suite.

Platform smoke tests:

See CONTRIBUTING.md for contribution rules and
SECURITY.md for private reporting.

Responsible use

is for authorized security assessment. Operators are responsible for
scope, approvals, target ownership, data handling, and local law.

Support

is an open-source project maintained by Strategic Automation Ltd.
If is useful to you or your organisation, you can
sponsor its continued de.

Sponsorship supports continued de, testing, documentation,
compatibility work, and releases.

remains a under the MIT licence. Sponsorship does not include
guaranteed support, feature priority, or influence o policy.

License

MIT — see LICENSE.

项目信息

开发语言：Python

Star 数：145

技术要点

该项目属于红蓝攻防方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕Supervised, Hermes-native penetration testing profile with 35 routed playbooks, guarded target execution, persistent engagement state, evidence-backed findings, and release-gated reporting.展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。