This commit is contained in:
yulinlin 2026-09-05 21:42:29 +02:00 committed by GitHub
commit 8105750358
No known key found for this signature in database
GPG key ID: B5690EEEBB952194

79
docs/safe-first-run.mdx Normal file
View file

@ -0,0 +1,79 @@
---
title: "Try Strix safely"
description: "A safe first-run example using a local, intentionally vulnerable demo app"
---
> **Safety first:** Only scan systems you own or are explicitly authorized to test. For your first Strix run, use a local demo target such as `127.0.0.1` or `localhost`.
## Why use a local demo target?
Strix is an autonomous penetration-testing tool. It can execute real requests, interact with the target, and validate findings with proofs-of-concept. A local, intentionally vulnerable app gives you a controlled environment to verify installation, credentials, scope, and output before testing anything real.
## 1. Start a local vulnerable app
This example uses OWASP Juice Shop, a well-known intentionally vulnerable web application.
```bash
docker run --rm --name strix-demo -p 3000:3000 bkimminich/juice-shop
```
Wait until the app is available at `http://127.0.0.1:3000`.
## 2. Run Strix against the local target
In another terminal:
```bash
export STRIX_LLM="openai/gpt-5.4"
export LLM_API_KEY="your-api-key"
strix --target http://127.0.0.1:3000
```
The first run may pull the sandbox Docker image. Results are saved under `strix_runs/<run-name>`.
## 3. What to expect
Strix will plan and execute tests against the local target, then summarize findings. A typical result includes:
- Confirmed or suspected vulnerabilities, grouped by severity
- Evidence and reproduction steps
- Proof-of-concept artifacts where validation succeeded
- Remediation guidance
Because Juice Shop is intentionally vulnerable, you should expect findings. Treat this run as a way to become familiar with Strix's output and review workflow.
## Minimal safe scope checklist
Before running Strix, confirm:
- The target is local-only, such as `http://127.0.0.1:3000`
- You own the target or have explicit permission to test it
- The target is isolated from production systems and real user data
- You are not using production credentials or secrets
- You can stop the container when finished
To stop the demo app:
```bash
docker stop strix-demo
```
## How Strix differs
| Dimension | Strix | DAST scanner | Template-based scanner | Manual pentest framework |
|---|---|---|---|---|
| Primary purpose | Authorized, agentic penetration testing | Web crawling and rule-based vulnerability detection | Known-issue/template matching | Human-driven exploitation and post-exploitation |
| Workflow | Plans and executes tests based on target and scope | Spider plus rule requests | Bulk probing for known patterns | Operator selects and runs modules |
| Typical output | Attack paths, evidence, reproduction steps, remediation guidance | Vulnerability list with request/response details | CVE or misconfiguration list | Exploit/session output |
| Safety controls | Target/scope constraints and local demo guidance | Scope and auth configuration | Target lists and rate limits | Operator discipline |
| Best first run | Local DVWA/Juice Shop-style target | Local or staging app | Local or staging app | Lab environment |
Strix is not an indiscriminate scanner. It is designed for authorized security testing, and the safest first run is always against a local, disposable target.
## Next steps
- Review the generated report and evidence before acting on findings
- Configure scope/target restrictions before testing a real application
- See [configuration](/advanced/configuration) for available runtime settings
- See [scan modes](/usage/scan-modes) if you want to control scan depth