scan-redteam
Routes giskard.scan vulnerability, quality, suite-generation, knowledge-base, prompt-injection, and third-party red-team scan tasks.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- scan-redteam
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Routes giskard.scan vulnerability, quality, suite-generation, knowledge-base, prompt-injection, and third-party red-team scan tasks.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/api-reference.md9.7 KB
- 参考文档references/third-party-integrations.md8.5 KB
- 参考文档references/troubleshooting.md10.8 KB
- 参考文档references/workflows.md9.5 KB
03 SKILL.md 正文
Scan and Red-Team Workflows
Use this sub-skill when the task is about giskard.scan: vulnerability scans,
RAG quality scans, generated scan suites, scan scenario generators, prompt
injection datasets, KnowledgeBase, or optional third-party scanners such as
garak, deepteam, and private lidar.
Read First
- API reference for verified scan entry points,
signatures, generator registries,
KnowledgeBase, and item listing. - Workflows for planning and running native Giskard scans safely, including target design, concurrency, knowledge bases, and default generator/provider prerequisites.
- Third-party integrations for
third_party_scan(tool=...), optional extras, supported item names, and skip behavior for garak, deepteam, and lidar. - Troubleshooting for missing credentials, empty knowledge bases, network/Hugging Face issues, commercial-use flags, target state safety, and optional scanner failures.
- scripts/inspect_scan_api.py for a safe
installed-package inspection helper. It imports
giskard.scan, prints public versions/items, builds a tinyKnowledgeBase, and checks optional scanner availability without running scans or providers.
Typical Triggers
- "Run or configure a
vulnerability_scanfor my chatbot." - "Plan a RAG
quality_scanwith a knowledge base." - "Generate a scan
Suitefrom specific scenario generators." - "List built-in scan generators or garak/deepteam probes."
- "Use prompt-injection or Hugging Face dataset scenarios."
- "Choose between native Giskard, garak, deepteam, and lidar scanning."
- "Explain why scan results are skipped, errored, or grouped unexpectedly."
Boundary and Routing
Include here:
vulnerability_scan,quality_scan,generate_suite, andDEFAULT_TARGET_MODE.KnowledgeBase,Document,KnowledgeBase.from_texts, lazy embeddings, and empty-knowledge-base behavior.ScenarioGenerator,SuiteGeneratorRegistry, built-in vulnerability and quality generator registries,PromptInjectionScenarioGenerator, andHuggingFaceDatasetScenarioGenerator.list_scan_items("giskard" | "garak" | "deepteam")andthird_party_scan(tool="garak" | "deepteam" | "lidar")prerequisites.- Scan execution controls:
max_scenarios,seed,group_by,parallel,max_concurrency,return_exception,target_mode, and vulnerability-onlycommercial_use.
Route elsewhere:
- Suite and check-result interpretation, custom checks, JUnit export, and
deterministic
Scenario/Suitedetails -> checks-evals. - Default generator/provider keys,
provider/modelrouting, provider extras, and live LLM errors -> llm-providers. - Designing async targets, agent workflows, tools, prompt templates, and stateful chat wrappers -> agents-workflows.
Operating Rule
Do not start an expensive or live red-team scan until the user has confirmed the
provider credentials, network/commercial-use stance, target concurrency safety,
and optional scanner extras needed by the selected workflow. For no-key
inspection, run only the bundled inspect_scan_api.py helper.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.