测试与安全技能

Run Isaac for Healthcare workflows in a simulator using a learned policy or rule-based controller. Verify completed episodes through the final success summary, visible behavior, and recorded data, with checkpoint compatibility checks when a checkpoint is supplied.

中文简介正在准备中。当前显示英语简介。

• Evaluating a learned policy for a specified number of episodes
• Checking a workflow with its declared rule-based controller
• Checking a supplied checkpoint and completing simulator evaluation

Guide checks of a customized Jetson system image, from file consistency to tests on a flashed device over SSH or UART. Device access is defined, while the broader validation procedure remains a skeleton.

中文简介正在准备中。当前显示英语简介。

• Check system image files without a connected device
• Run smoke tests after flashing a Jetson device
• Investigate kernel and module version mismatches

Add and run unit or functional tests for Megatron-LM. The guide explains test organization, YAML execution recipes, reference results for numerical checks, test filters, temporary disabling, and reproducing continuous integration failures locally.

中文简介正在准备中。当前显示英语简介。

• Add unit tests or functional training tests
• Run selected tests and reproduce CI failures locally
• Download or update reference results for numerical checks

Measure video encoding or decoding throughput on Jetson, compare presets, and test concurrent worker capacity. When representative media is unavailable, produce a clearly labeled estimate from SDK documentation, keeping estimates separate from measured results.

中文简介正在准备中。当前显示英语简介。

• Measure encoding or decoding speed with representative media
• Compare throughput between P4 and P5 presets
• Test capacity as concurrent worker counts increase

Run quality assurance tests on Holoscan Sensor Bridge hardware from a supplied test document. Select tests that match the connected equipment, execute automatable cases, evaluate pass or fail, and produce a structured results report.

中文简介正在准备中。当前显示英语简介。

• Run a documented test plan on an HSB board
• Select automated tests for the connected sensors
• Investigate failed tests using execution logs

Build and run a benchmark for RDMA WRITE, a direct memory transfer initiated by a CUDA kernel. Check GPU and network adapter pairing, interpret timing columns, and assess median latency, p99, and timing variation.

中文简介正在准备中。当前显示英语简介。

• Measure GPU-initiated RDMA WRITE latency
• Assess p99 latency and jitter for a control loop
• Compare results before and after hardware tuning

Capture and decode execution traces from BlueField DPA workloads to inspect kernel events, synchronization, and communication timing. Choose capture modes, adjust file limits and recording overhead, and diagnose empty traces or decoding failures.

中文简介正在准备中。当前显示英语简介。

• Investigate incorrect results from a DPA kernel
• Inspect delays between kernel entry and communication
• Adjust capture modes and overhead

Measure DOCA Flow rule update and disable rates when work runs on the device's DPA processor. Configure workloads, interpret repeated measurements in thousands of operations per second, and investigate zero-operation results or failed self-tests.

中文简介正在准备中。当前显示英语简介。

• Establish a DPA rule-update performance baseline
• Compare burst sizes, queue sizes, and worker counts
• Review iteration statistics and measurement conditions

Deploy and operate DOCA Argus Service on BlueField to monitor suspicious activity. This skill guides detection settings, forwarding findings to a security monitoring system, and troubleshooting missing alerts, excessive noise, or performance impact.

中文简介正在准备中。当前显示英语简介。

• Deploying runtime security on BlueField
• Forwarding security findings to Splunk
• Investigating missing or excessive alerts

Test a running AMC service with the bundled four-camera warehouse dataset. Run camera calibration from upload to completion, then check distance and image projection errors against ground-truth data to verify the installation.

中文简介正在准备中。当前显示英语简介。

• Verify a newly installed AMC service
• Check the sample's calibration errors against ground truth
• Walk through the sample workflow in Swagger UI

Check one DICOM medical imaging series before conversion or AI inference by inspecting headers only. Produce a JSON report with file inventory, image orientation, personal information flags, findings, and a pass, warn, or fail verdict.

中文简介正在准备中。当前显示英语简介。

• Checking a DICOM series before conversion
• Reviewing image orientation before AI inference
• Recording preflight findings in JSON

Check a live app or site before launch for exposed secrets, database access risks, mobile layout, search visibility, and speed. The skill produces scores, a launch verdict, and a prioritized list of fixes.

中文简介正在准备中。当前显示英语简介。

• Checking whether a live app is ready to launch
• Looking for API keys exposed in frontend code
• Checking Supabase or Firebase access settings

Audit saved agent sessions for access to private files and evidence that secret values entered an agent's context. Get findings without displaying secrets, identify keys needing rotation, and inspect Claude Code or Codex sessions in a local report.

中文简介正在准备中。当前显示英语简介。

• Check whether an agent accessed .env files
• Identify keys that may need rotation
• Audit private-file access by subagents

Audit a completed AI development phase to check whether its planned evaluation strategy was implemented. Rate coverage for each evaluation dimension and produce an EVAL-REVIEW.md report with scores, gaps, a verdict, and a remediation plan.

中文简介正在准备中。当前显示英语简介。

• Review evaluation coverage after an AI development phase
• Check implementation against the AI-SPEC.md evaluation strategy
• Identify partially covered or missing evaluation dimensions

Review acceptance tests and verification records across development phases to find unfinished checks. Compare records with the code to identify outdated documentation, then produce a prioritized plan for tests that people need to perform.

中文简介正在准备中。当前显示英语简介。

• Find unfinished acceptance tests across phases
• Review skipped, blocked, and human-dependent checks
• Identify test documentation that no longer matches the code

Add unit and end-to-end browser tests to a completed development phase using its acceptance criteria, planning documents, and implementation. Review the proposed coverage, verify tests fail and then pass, and commit the generated test files.

中文简介正在准备中。当前显示英语简介。

• Add tests after completing a development phase
• Plan test coverage from acceptance criteria
• Focus additional tests on module edge cases

Run an audit, separate automatically fixable findings from issues requiring manual work, and carry eligible fixes through testing and individual commits. Limit processing by severity or count, or preview classifications without changing files.

中文简介正在准备中。当前显示英语简介。

• Taking audit findings through fixes, tests, and commits
• Previewing which findings can be fixed automatically
• Processing only high-severity findings

Test text watermark removal tools using samples marked with a known secret key. Measure whether the mark survives, alongside changes to meaning, facts, wording, and length, and compare recorded runs in a shared table.

中文简介正在准备中。当前显示英语简介。

• Check a humanizer before paying for it
• Compare watermark removal tools using the same measurements
• Measure meaning and fact changes after rewriting

Run CIAgent regression checks after changing an agent’s code, prompts, or knowledge base. Interpret failures and repeated runs to distinguish incorrect answers, inconsistent model judgments, and setup problems before committing changes.

中文简介正在准备中。当前显示英语简介。

• Check agent behavior after prompt changes
• Review checks after knowledge base updates
• Investigate inconsistent test results

Test local web applications with Python and Playwright, a browser automation library. Verify interface behavior, investigate problems, capture screenshots, and collect browser logs, with a helper script that manages one or more application servers.

中文简介正在准备中。当前显示英语简介。

• Checking button and form behavior in local web applications
• Investigating interface problems with screenshots and logs
• Automating browser checks for static HTML pages

Guide software testing with a test-first workflow, from small unit checks to complete application flows. Apply consistent test structure across supported frameworks, keep tests independent, and use meaningful checks when measuring coverage.

中文简介正在准备中。当前显示英语简介。

• Writing tests before implementing a feature
• Organizing unit, integration, and end-to-end tests
• Reviewing mocks and test independence

Set up CIAgent tests that check whether changes affect an AI agent’s behavior. Create a runner, record reference answers, add factual checks, and generate reports that distinguish agent variability from unstable evaluations.

中文简介正在准备中。当前显示英语简介。

• Adding regression tests to an existing AI agent
• Checking factual answers and out-of-scope responses
• Measuring answer stability across repeated runs

Analyze audio, images, video, and text for AI generation or manipulation with Resemble AI. Receive detection scores with explanations, trace synthetic audio sources, apply or detect invisible watermarks, and compare speakers against registered voice profiles.

中文简介正在准备中。当前显示英语简介。

• Check audio, images, or video for AI generation
• Identify the platform behind synthetic audio
• Apply or detect invisible media watermarks

Check an agent’s test claims against local Claude Code transcripts and the git working tree. Nine rule-based checks identify gaps such as missing runs or failed tests, with timestamps and quoted evidence, without calling a model.

中文简介正在准备中。当前显示英语简介。

• Verify that reported passing tests actually ran
• Investigate when and why a test suite became smaller
• Review a completed session before trusting its summary