Works with the AI Agents you already use
Web Penetration Testing
Chain web findings into a proven intrusion, only as far as the objective needs
$24
Web Penetration Testing
Example session with this skill installed
Authorized web intrusion test. Objective: prove an unauthenticated user can reach the admin dashboard. Stop when proven.
- Read your context and instructions
- Compiled the web penetration testing
PATH-ID: WEB-01
AUTHORIZED TARGET: staging app <SYNTHETIC_TARGET>; unauth start.
START CONDITION
No authenticated session.
FINDINGS USED
Forced-browsing lead — admin route is not linked but is guessable.
ATTACK STEPS
- GET /admin → redirects to /login, showing a page-level guard.
- GET /api/admin/stats directly → synthetic example returns admin data without authentication.
ACHIEVED ACCESS
Admin API data reachable without authentication.
IMPACT DEMONSTRATED
Broken function-level authorization on the administrative backend path.
STOP CONDITION
Objective proven at step 2. Testing stops; no further enumeration is performed.
CLEANUP / ROLLBACK
None required in this synthetic read-only example.
EVIDENCE
Illustrative request/response path only. No external system was contacted.
RESIDUAL RISK
Other administrative endpoints may share the same authorization gap; a bounded variant search is recommended after remediation.
Connects securely to your tools. The creator never sees your data.
About this skill
An authorized, objective-driven web application penetration test that chains only the findings needed to prove an agreed impact. It covers browser-facing attack paths such as authentication bypass, unsafe state changes, cross-tenant access, SSRF, file handling and injection while enforcing an explicit authorization gate and a hard stop condition.
What it does
- Objective-driven web penetration testing against an explicitly authorized application and scope.
- Authentication and session boundary testing for bypasses and unsafe browser flows.
- State-change and access-control testing across user roles and tenant boundaries.
- Web attack-path validation for SSRF, file handling, injection and related reachable weaknesses.
- Minimal exploit chaining only when necessary to prove the agreed objective.
- Evidence-backed reporting that separates observed runtime evidence from hypotheses and missing inputs.
How it works
- Authorization and scope gate confirms the target, permitted actions, starting identity and stop condition.
- Baseline capture establishes expected behavior through the real web/browser surface.
- Boundary probes test authentication, authorization, state transitions and relevant input paths.
- Minimal chaining combines findings only when required to demonstrate the agreed impact.
- Stop condition ends active testing as soon as the objective is proven.
- Evidence record documents the reproducible path, impact, limitations and recommended follow-up.
Boundaries
API-only objectives route to api-penetration-testing. Identity-only privilege or tenant-boundary objectives route to identity-penetration-testing. Single documented findings that only need exploitability confirmation route to exploit-validation.
The skill does not perform open-ended destructive testing, denial-of-service activity, attacks against real users, or work outside explicit authorization and scope.
Why this beats prompting it yourself
Generic penetration-testing prompts can drift into broad scanning, overclaim impact, or continue after the objective is already proven. This skill imposes an authorization gate, explicit objective, evidence discipline and a hard stop condition so the agent performs the smallest sufficient test instead of turning every web issue into an uncontrolled campaign.
Validation
Routing boundary stress-tested in isolated Antigravity sessions — 5/5 high-risk negative cases kept this skill OFF with zero false triggers. Independent isolated routing cases also selected web-penetration-testing correctly for browser-flow pentest and authorized web exploit-chain requests.
Use cases
- Pentest a web application for browser-based authentication bypass.
- Prove an unauthorized state change through the real web flow.
- Test whether a browser-facing weakness can be chained to the agreed impact.
- Validate cross-tenant or role-boundary behavior exposed through the web application.
- Demonstrate a bounded web intrusion objective and stop once it is proven.
Known limitations
Requires explicit user-supplied authorization, target scope and engagement boundaries before active execution. Runtime claims require observable browser/network evidence; unavailable evidence remains BLOCKED or UNVERIFIED rather than being invented.
Part of JetFerrari Security Engineering.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
Active penetration testing requires explicit authorization, defined scope, and a clear testing objective. The workflow focuses on web application attack paths and stops when the agreed objective has been sufficiently demonstrated. API only objectives, identity specific testing, broad vulnerability discovery, and unrelated infrastructure assessment belong to their dedicated workflows.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 25 days ago
Permissions
Project access is task-scoped; no fixed file path or external account is required. Do not grant broader authority than the current task. Active testing requires explicit user-supplied authorization and scope. Credentials/tokens must be supplied for the authorized target; the skill does not harvest environment variables.