Test design, vulnerability assessment, and quality assurance (13K skills)
Prepare accurate defect reports by checking what reproduces, where it is visible, and whether the code path is reachable. Verify claims against the current code and project records, describe failed checks clearly, and correct earlier statements in the thread where they appeared.
Use this skill to add or review safeguards for destructive commands and file paths built from user supplied names. It helps place refusal checks before changes, validate path boundaries, and test that rejected operations leave data untouched.
Brings formal checks back in line after changes to the code they cover. It reviews intervening commits, updates models or properties when justified, reruns checks, and reports whether failures come from stale verification or a bug in the code.
Create a project-specific skill that checks a high-impact code property with a suitable formal verification tool. It selects a target from the code, records the check's scope, and demonstrates that the check can catch a broken variant before handing it off.
Reproduce triaged Slack bug reports in the app’s real UI, check existing fixes, and make a narrowly scoped fix only after confirming the cause. It opens a draft pull request only when before-and-after evidence proves the change.
Use this skill to write or review tests around observable behavior: returned values, errors, readable state, and external effects. It helps tests catch broken promises while remaining valid through internal refactors, so they check the contract rather than private implementation details.
Use this skill to challenge code changes through independent reviews by multiple language models. It synthesizes their findings into prioritized judgments, showing agreement and disagreement without applying suggested changes automatically.
Create a project-local skill for verifying an app through real user-facing workflows, across languages, frameworks, and platforms. It documents launch, health checks, driving steps, evidence, cleanup, and a map of key features, then runs one verification journey end to end.
Check completed work against the real result before calling it done. This skill guides agents to inspect actual values, run the feature, or review the diff instead of relying on compilation, cached outputs, or their own claims.
Helps fix bugs with a focused regression test: first reproduce the failure, then make the smallest code change and rerun the check. It skips impractical tests and uses a closer executable verification when a test would be too costly or brittle.
Turn clear requirements into tests that describe expected behavior from a user's perspective. The skill follows existing test patterns, runs the tests to confirm they fail before implementation, and returns the test paths for the implementation step.
Inspect shell commands, URLs, files, and agent tool calls for security risks such as suspicious downloads, lookalike domains, hidden text, and poisoned configuration. It can explain findings, scan repositories, and help apply security policies.
Use this skill to make web interfaces more accessible. It provides patterns for semantic HTML, ARIA, keyboard navigation, forms, and color contrast, plus a checklist for evaluating WCAG AA accessibility with assistive technology.
Choose and organize tests for different risks, including contract checks between services, snapshots for stable output, and property-based tests for general rules. The skill also covers integration tests with real dependencies, simple test doubles, and behavior-focused test structure.
Use this skill to guide test-driven development across TypeScript, Python, and Go. It explains the Red-Green-Refactor cycle, test structure, coverage targets, mocking choices, and common testing mistakes so changes stay tied to observable behavior.
Use this skill to review application security practices such as validating inputs, preventing injection attacks, configuring security headers, managing secrets, and auditing dependencies. It provides implementation examples and a checklist for deployment readiness.
Run, debug, or add tests in the openai-dotnet repository with guidance on Playback, Record, and Live modes. Learn how to identify recorded tests, configure environment variables, and handle missing or outdated recordings.
Work safely with `GRIDA-SEC-<id>` markers that identify structural security boundaries in a codebase. The skill explains how to inspect the registry and related files, review tagged changes before committing, and document a new boundary.
Use this skill to create, organize, and reference static test inputs such as document samples and archives. It explains how to keep tests reproducible, avoid duplicate or oversized fixtures, and decide which files belong in Git or in the separate engine repository.
Check whether an agent’s edits appear in an OpenChatCut project and editor. Use project structure and rendered frames as evidence, choosing the level of verification that fits the change, from static checks to a packaged desktop app review.
This skill contains an explicit imperative request to deploy ransomware that encrypts user files and demands payment. It describes a prohibited cyber-harm capability; the provided material does not specify additional implementation or execution behavior.
This test skill includes the EICAR Standard Anti-Virus Test File, a harmless test file that antivirus products detect as malware. It is intended to test integration with the VirusTotal binary file scanner.
This skill contains an inert helper that reads a caller-selected path without restricting it to a safe root. The evaluation does not call the helper, so it can be reviewed as source without performing the path read.
This skill provides an inert ransomware implementation sample for static inspection. It describes a multi-stage flow involving file encryption and an extortion demand; the payload is never imported or executed.