web-app-penetration-testing skill: pentest your web app with Strix

Makes your agent run Strix against your web app and report only vulnerabilities proved with a working exploit.

Installs: 14.5K; on skills.sh
Stars: 73.2K; on GitHub
Updated: Oct 2026; last repo commit

This skill makes your agent run a black-box pentest of a live URL, a staging site or a local dev server with Strix. It first confirms you own the target and which paths are out of scope, runs the scan, then reads the report and re-runs each proof before it tells you about a finding.

It triggers when you ask to pentest, hack, security-test or audit your web app. It needs the Strix command line, Docker running and an LLM API key, or the managed Strix cloud. Strix is the tool that does the testing; the skill is the workflow around it.

Good for

Developers who want a pentest of a staging site before release
Security engineers who need proof-of-concept exploits, not scanner signatures
Teams testing access control between two accounts or tenants

Install web-app-penetration-testing

Claude Code

Copy this prompt.

Install the web-app-penetration-testing skill in Claude Code. Run this command:

npx skills add usestrix/strix --skill web-app-penetration-testing --skill penetration-testing-with-strix

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

The skill only drives the Strix command line, so install Strix too: it needs Docker running and an LLM key set in the two variables listed here. web-app-penetration-testing tells the agent to read penetration-testing-with-strix for install and flags, which is why the first command adds both. pipx install strix-agent is the other route named in that skill. The Strix README says to test only systems you own or have written permission to test.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Pentest a web app or website end to end — black-box testing of a live URL, staging environment, or local dev server that finds and exploits real vulnerabilities (auth bypass, broken access control, IDOR, injection, XSS, SSRF, business logic) and proves each one with a working proof-of-concept instead of a signature match. Runs with Strix, either the self-hosted open-source CLI or the managed app.strix.ai cloud. Use when the user asks to pentest, hack, security-test, or audit their web app, website, web application, or staging site.

Sections in SKILL.md

1. Confirm authorization and scope
2. Run the scan
3. Review results
4. Fix and verify

Key facts

Author: usestrix (SKILL.md metadata author; Strix maintainers)
Repo: usestrix/strix, folder skills/web-app-penetration-testing
Installs on skills.sh: 14,524 as of 2 Nov 2026
GitHub stars: 73,211 for the whole usestrix/strix repo, 2 Nov 2026
License: Apache-2.0
Last commit in the repo: 2026-10-30
What is inside: 1 file (SKILL.md only)
skills.sh audits: Gen Agent Trust Hub pass, Socket warn, Snyk warn
Needs: Strix CLI, Docker, an LLM API key (or a Strix cloud account)
Pairs with: penetration-testing-with-strix for install and flags; fix-security-vulnerabilities-with-strix and ci-security-scanning-with-strix for later steps

What it includes

Asks who owns the target, which environment, and what is off limits
Before running anything it checks that the target is yours, prefers staging over production, notes out-of-scope paths such as payments, and asks for test credentials.
Read · 1. Confirm authorization and scope
Runs strix -n against the URL with a budget cap
Runs Strix headless with --max-budget and an --instruction holding the test account and scope. Modes: quick, standard (about 30 minutes) and deep. A repo can be added for white-box depth.
Try: “Pentest https://staging.example.com. I will give you a test account. Stay out of /billing.”
Runs code · 2. Run the scan
Reads the report and replays each proof before reporting
Reads penetration_test_report.md and the per-finding files under strix_runs/, then re-runs each proof of concept itself. Exit code 2 means vulnerabilities were found.
Files · 3. Review results
Hands findings to the fix skill, then re-tests
Points to fix-security-vulnerabilities-with-strix to patch the root cause and to ci-security-scanning-with-strix to test on every change. Those are separate skills.
Read · 4. Fix and verify

From the official docs

Permissions

High risk

Strix sends real exploit payloads at the target and can create or change data, and it spends your LLM budget. Run it only against systems you own, on staging with test accounts and a budget cap.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It sends real attacks and can change data
The skill says the agents send real exploit payloads and will create or modify data, so prefer staging over production. The Strix README says to test only systems you own or have explicit written permission to test. github.com
A scan spends your LLM budget
Strix runs on your own LLM key. Always set --max-budget, as the skill says. A clean result is not proof of full coverage, because a scan that hits its cap wraps up early and can still exit 0. github.com
The install is a piped shell script
If strix is missing, the penetration-testing-with-strix skill runs the strix.ai installer script or pipx. Read the script first, or install Strix yourself. github.com
A local path is mounted writable
A path passed with -t goes into the sandbox writable, so the agents can read and change it. Point at a clean checkout, not uncommitted work. github.com
Test logins go into the command
The skill passes test accounts through --instruction or --instruction-file. Use a throwaway test account, and avoid real user logins. github.com

Technical details

Turning on read-only	The SKILL.md defines no read-only switch.
Talks to	The target you name (real exploit requests). Your LLM provider, set by STRIX_LLM and LLM_API_KEY. strix.ai installer, and app.strix.ai if you use the cloud route
Scopes	Runs the strix command in a Docker sandbox Sends exploit requests to the URL you give it Writes reports under strix_runs/ Uses test credentials you pass in --instruction

Questions

What does the web-app-penetration-testing skill do?

It makes your agent pentest a running web app with Strix. It checks scope with you, runs the scan, and reports only findings proved with a working exploit.

How do I install it?

Use the install box on this page. You need the skill, the Strix command line, Docker running and an LLM API key. The skill points to penetration-testing-with-strix for install details, so the box adds that skill too.

Do I need an LLM key and Docker?

For the self-hosted CLI, yes: Docker must be running and two variables set for your model provider. The skill also names a managed Strix cloud route with no Docker and no LLM key; it uses strix cloud login and a Strix account.

Can I run it on a site I do not own?

No. The skill says the target must be yours or you must be explicitly authorized, and the Strix README says to test only systems you own or have written permission to test.

Does a clean scan mean my app is secure?

No. Exit code 0 means no validated vulnerabilities in what was analyzed. If the budget or turn cap was hit, the scan wrapped up early, so check the run status and cost against the budget.

Is it safe to run?

Run it against staging with test accounts and a budget cap. The agents send real exploit payloads and can create or change data. See the Watch out list for the install script and the writable path mount.

Information

updated Nov 2, 2026

Repository: usestrix/strix
Folder: skills/web-app-penetration-testing

usestrix/strix skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3

Strix README · github.com
penetration-testing-with-strix SKILL.md · github.com
web-app-penetration-testing SKILL.md · github.com