Skip to content

jaktestowac/awesome-copilot-for-testers / plugins/testing-application-security/skills/testing-application-security

testing-application-security

Unverified

Applies OWASP-informed security testing within a tester scope: an authorization gate before any probing, authorization matrices, broken access control and IDOR checks, input validation and injection observation, session and auth behaviour, and safe reporting of findings. Use when testing auth-protected features, when a security review needs QA coverage, when an exploratory session surfaced something security-shaped, or when access control needs a systematic check rather than a spot check.

resources/scope-and-authorization-gate.md

View on GitHub
# Scope and Authorization Gate

Complete this before any probing. Not a formality: unauthorized testing is unauthorized access, regardless of intent, employment, or how obviously the system belongs to your company.

---

# Authorization record

- **System**: [name and exact hosts or URLs]
- **Environment**: [staging / preprod / **production**]
- **Authorized by**: [name, role]
- **Authorization form**: [email, ticket, signed statement, bug bounty scope page] with [link]
- **Date obtained**: [YYYY-MM-DD]
- **Window**: [start] to [end]
- **Tester**: [name]

## In scope

- Hosts: [exact list; wildcards only if the authorization uses them]
- Endpoints: [paths or "all paths on the listed hosts"]
- Accounts: [the synthetic accounts created for this]
- Roles: [which role boundaries are being checked]

Everything not listed is out of scope. A related host that "is obviously ours" is not in scope unless it is named.

## Out of scope, explicitly

- Third-party services, including the payment provider, the email service, the CDN, the auth provider
- Any host not listed above
- Any account belonging to a real person
- Production data of any kind

## Prohibited actions

- Denial of service, load generation, or resource exhaustion
- Data exfiltration beyond the minimum needed to demonstrate a finding
- Modifying or deleting data belonging to anyone else
- Social engineering, phishing, or any action involving a person rather than a system
- Physical access
- Persistence: no backdoors, no accounts created for later use
- Testing outside the authorized window

## Contact

- **Immediate contact**: [name, phone, channel]
- **When to contact them**: any unexpected outcome, any sign that real data is involved, any finding above medium severity, any action with an effect you did not predict

## Data rule

Synthetic accounts and synthetic data only. If real customer data appears at any point, **stop, do not read further, and contact the person above**. See `handling-sensitive-test-data`.

---

## What valid authorization has to state

A message saying "sure, go ahead" is not enough. The record needs:

| Element | Why |
| --- | --- |
| Named system and hosts | So scope is not inferred |
| Environment | Production carries different risk and needs its own approval |
| Named authorizer with the authority to give it | A colleague's agreement is not authorization |
| Time window | Open-ended authorization is not authorization for a year from now |
| Prohibited actions | So the boundary is written rather than assumed |
| A contact | So an unexpected outcome has an immediate path |

## Production

Production testing needs its own explicit approval, separate from any staging authorization, and even then:

- **no probing against real customer accounts**, only accounts you created
- **no destructive or state-changing actions** outside your own test accounts
- **no load generation** of any kind
- **someone watching monitoring** during the window, aware that the traffic is yours
- **an agreed abort signal** and a way to reach you instantly

Default to staging. The usual reason given for production, that staging differs, is a finding about environment parity worth reporting on its own.

## Third-party systems

- A published **bug bounty scope** is authorization for exactly what it lists. Read the exclusions; they are the part that matters.
- **No scope statement means no authorization.** A vendor's system being reachable from your test environment does not put it in scope.
- Testing an integration means testing **your side** of it. Stub the third party.

## Stop conditions

Stop immediately, and contact the person above, when:

- real customer data appears anywhere in a response
- an action has an effect you did not predict
- a service degrades or errors in a way that suggests your traffic caused it
- a finding reaches high severity, where continuing means exploiting rather than demonstrating
- you reach the edge of the written scope, even by one host
- the time window ends
- you are uncertain whether the next action is authorized

The last one is the most important. Uncertainty resolves by asking, not by proceeding and explaining afterwards.

## Depth limit

This skill covers verification that the application enforces what it claims. It stops at:

- confirming a class of finding with the minimum demonstration
- reporting it through the disclosure path

It does not cover, and a tester operating under this skill should hand off rather than attempt:

- developing a proof of concept into a working exploit
- chaining findings into a fuller compromise
- post-exploitation of any kind
- testing infrastructure, networks, or hosts rather than the application
- anything involving a person rather than a system

Handing off is the right outcome, not a failure. A confirmed IDOR reported clearly is more useful to a security team than a half-built exploit.

## If something goes wrong

1. **Stop.** No further requests.
2. **Contact** the named person immediately. Speed matters more than a complete account.
3. **Preserve** what you have: requests, responses, timestamps. Do not clean up.
4. **Do not attempt to fix it yourself**, particularly if data was modified.
5. **Write down** what you did, in order, while you remember it precisely.

Reporting an accident promptly is always better than the alternative, and every incident process assumes it.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall