Test design, vulnerability assessment, and quality assurance (13K skills)
Design and write Python tests with pytest, using test-driven development (writing a failing test before implementation), reusable fixtures, mocks for external dependencies, parameterized cases, and coverage reports. It also covers async code and organizing unit and integration tests.
Review medical device technical files against EU MDR 2017/745 requirements. Check for documents such as clinical evaluation reports and post-market surveillance plans, then organize missing or incomplete items into a structured audit report with compliance findings.
Review an agent skill before release by checking its structure, security, written quality, and behavior on generated test cases. It provides scores, review materials, prioritized fixes, and an improved SKILL.md based on the findings.
Validates performance hypotheses through controlled experiments. It checks a clean baseline, applies one hypothesis-driven change, runs at least two validation passes, and records measured differences and evidence before restoring the baseline for the next experiment.
Verify XQUIC code, build, and test changes with required happy-path and error-path unit tests, the full local unit suite, and endpoint-visible checks for production behavior. Receive concrete test evidence and clear reporting of any coverage gaps before a commit or pull request.
Pre-review a published XQUIC code pull request against its exact base-to-head changes. The review checks protocol conformance, test coverage, adversarial cases, hot-path performance, and memory safety, then records evidence and a gate result in a local report.
Review UI files against Web Interface Guidelines to find design, UX, and accessibility issues. The skill reports findings in a concise file-and-line format, helping you locate areas that may need changes. It fetches the latest guidelines before each review.
Use browser commands to navigate websites, interact with page elements, fill forms, and inspect results. The skill can also capture screenshots or PDFs, record videos, and extract page information for testing and documentation workflows.
Turn the current branch diff or a specified pull request into an actionable testing plan. It summarizes the changes, assesses risk, and selects relevant checks for critical behavior, regression, edge cases, accessibility, and visual changes so a team can review PR readiness.
Run authorized security assessments through YAML workflows that coordinate AI agents and security tools. The skill supports reconnaissance, vulnerability scanning, analysis, and reports in Markdown, HTML, or JSON, with findings linked to captured execution evidence.
Create a test client for an MCP server, a program that exposes tools to clients. The examples launch the server over standard input and output, connect through the MCP SDK, call a named tool with arguments, and close the process after testing.
Set up snapshot tests for command-line output. The skill provides a pattern that saves expected output, compares future runs against it, and supports updating snapshots when output changes, helping teams review differences in CLI behavior.
Set up an end-to-end test harness for command-line applications. It launches a CLI process, captures standard output and error, records its exit code, and provides assertions for expected output, snapshots, and exit status.
Compare algorithm implementations using shared test cases to check outputs, benchmark runtime and memory, and stress-test with random inputs. When solutions disagree, narrow the discrepancy to a minimal failing case for easier debugging.
Create a BATS test scaffold for shell scripts, including test files, fixtures, setup and teardown routines, assertions, and mock helpers. It helps organize isolated tests for command behavior, input handling, expected output, and generated files.
Generate edge cases, random inputs, and large stress tests to check algorithm solutions. Compare results with a brute-force reference, then find and minimize counterexamples when outputs differ, making failures easier to reproduce and debug.
Create mock standard input streams for testing interactive command line tools. The skill provides patterns for supplying scripted answers to a command, running it with temporary mocked input, and restoring the original input afterward.
Find personally identifiable information in conversational or other text and choose ways to protect it, such as masking, pseudonymization, tokenization, or encryption. It also covers configurable detection, compliance logging, and audit trails for handling sensitive data.
Check an MCP App before deployment with a structured verification workflow. It covers runtime behavior, handler order, text fallbacks, resource links, single-file builds, host styling, content security policy, and leftover migration patterns.
Audit an MCP App running in a sandboxed iframe by finding its network destinations in build output and source code. Classify origins for resources, connections, and nested frames, then prepare matching Content Security Policy settings for `registerAppResource`.
Use this skill to detect prompt injection attempts in LLM applications and plan layered defenses. It covers input sanitization, detection methods such as classifiers and canary tokens, output validation, logging, and alerting.
Guide feature and bug fix work through test-driven development: write a failing test first, add the simplest code that makes it pass, then refactor while keeping tests green. It emphasizes verifying each stage before moving on.
Verify claims that work is complete, fixed, or passing by running the full relevant command, reading its output and exit code, and checking that the evidence supports the claim before reporting it or creating a commit or pull request.
Review an implementation against coding, test coverage, specification, performance, and security criteria, along with project-specific checks. The skill assigns a weighted score and gives actionable fixes for failed items, then can recheck after remediation.