Rai Standards
MicrosoftOfficialPublished by Microsoft
- Skill
- Guidelines
- Business and productivity
- CC-BY-4.0
- No issues found
AutoGineer summary
Written by AutoGineer from the skill’s own files, not by the author
The rai-standards skill provides a structured repository of reference material for Responsible AI (RAI) planning. It consolidates key frameworks such as the NIST AI Risk Management Framework (RMF) 1.0, an AI-specific adaptation of the STRIDE threat-modeling methodology, and a paraphrased overview of EU AI Act risk tiers. Designed to support AI planners, this skill maps specific RAI phases to relevant standards, ensuring that governance, risk management, and security analysis are grounded in established industry and regulatory references.
What it does
- Provides a centralized index of NIST AI RMF trustworthiness characteristics.
- Maps RAI project phases to corresponding standards and governance frameworks.
- Offers an AI-specific extension of the STRIDE threat-modeling methodology.
- Summarizes EU AI Act risk tiers and high-risk obligation themes.
- Maintains a catalog of open-standard resources for AI management and risk assessment.
- Defines a protocol for layering custom frameworks onto the default NIST baseline.
When to use it
- When scoping an AI project and identifying necessary governance frameworks.
- During the risk classification phase to align with regulatory tiers.
- When performing security model analysis using AI-adapted threat modeling.
- When documenting risk management, data governance, and human oversight requirements.
- When integrating sector-specific or regional AI standards into an existing RAI workflow.
Advantages
- Centralizes disparate RAI standards into a single, reusable package.
- Facilitates consistent mapping of project phases to regulatory and industry standards.
- Extends traditional security threat modeling to address AI-specific risks like data poisoning and model tampering.
- Supports interoperability by allowing custom frameworks to be layered over a NIST baseline.
- Provides clear attribution and licensing guidance for all included reference materials.
Usage examples
- “What are the NIST AI RMF trustworthiness characteristics?”
- The agent lists the seven characteristics including Valid and Reliable, Safe, Secure and Resilient, Accountable and Transparent, Explainable and Interpretable, Privacy-Enhanced, and Fair with Harmful Bias Managed.
- “How should I approach threat modeling for an AI inference endpoint?”
- The agent provides the AI STRIDE matrix, highlighting concerns like inference exfiltration, data poisoning, and identity confusion, and suggests treating the endpoint as a trust boundary.
Install
Download the skill as a ZIP:
https://mp-staging.autogineer.com/agent-skills/microsoft/rai-standards/downloadUnzip it into the cross-client skills folder of your project:
mkdir -p .agents/skills && unzip rai-standards.zip -d .agents/skills
Documentation by Microsoft
RAI Standards Skill
This skill is the reusable standards package for the RAI Planner. It consolidates the embedded NIST AI RMF content, the AI STRIDE threat-modeling overlay, and a paraphrased EU AI Act reference so the phase playbook can stay focused on workflow and orchestration.
Attribution and licensing posture
NIST AI RMF 1.0 is a U.S. Government document and is reproduced here as public-domain reference material with attribution. Public-domain material carries no license obligation, so it adds no term to the package license expression.
EU AI Act content in this skill is paraphrased and attributed rather than quoted verbatim, consistent with the open legal-text posture used in the repository. The paraphrase is this repository's own expression.
The AI STRIDE overlay is Microsoft-authored reference material for threat-modeling reuse in the RAI workflow.
Every file in this package is therefore governed by the repository-original license. The table records the source each file draws on.
| Path | License | Origin |
|---|---|---|
references/nist-ai-rmf-govern.md | CC-BY-4.0 | Adapted from NIST AI RMF 1.0 (public domain) |
references/nist-ai-rmf-map.md | CC-BY-4.0 | Adapted from NIST AI RMF 1.0 (public domain) |
references/nist-ai-rmf-measure.md | CC-BY-4.0 | Adapted from NIST AI RMF 1.0 (public domain) |
references/nist-ai-rmf-manage.md | CC-BY-4.0 | Adapted from NIST AI RMF 1.0 (public domain) |
references/eu-ai-act.md | CC-BY-4.0 | Paraphrase of EU AI Act 2024/1689, not reproduced |
references/ai-stride-overlay.md | CC-BY-4.0 | Repository-original |
SKILL.md and remaining package content | CC-BY-4.0 | Repository-original |
Framework index
NIST AI RMF trustworthiness characteristics
| Key | Characteristic | Description |
|---|---|---|
| validReliable | Valid and Reliable | Base characteristic for correctness, robustness, and stability |
| safe | Safe | Safety and harm prevention under normal and adversarial conditions |
| secureResilient | Secure and Resilient | Resistance to attack, misuse, and failure |
| accountableTransparent | Accountable and Transparent | Governance, auditability, and decision provenance |
| explainableInterpretable | Explainable and Interpretable | Understandability of model behavior and outputs |
| privacyEnhanced | Privacy-Enhanced | Protection of personal data and confidentiality |
| fairBiasManaged | Fair with Harmful Bias Managed | Bias detection, mitigation, and equitable outcomes |
Phase-to-framework mapping
| RAI phase | Primary standards package | Notes |
|---|---|---|
| Phase 1 Scoping | NIST AI RMF Govern + Map | Context, purpose, stakeholders, and policy framing |
| Phase 2 Risk Classification | NIST AI RMF Govern | Governance culture, DEI&A, and stakeholder engagement |
| Phase 3 Standards Mapping | NIST AI RMF Govern + Measure | Core standards mapping and TEVV alignment |
| Phase 4 Security Model Analysis | AI STRIDE overlay + Measure | Threat modeling and overlap with security analysis |
| Phase 5 Impact Assessment | NIST AI RMF Manage | Risk prioritization, mitigation, and monitoring |
| Phase 6 Review and Handoff | NIST AI RMF Manage + EU AI Act | Regulatory review, incident response, and evidence handoff |
Customer extension pattern
The default framework remains NIST AI RMF 1.0. When a customer supplies an additional framework, preserve the default NIST mapping as a baseline and layer the custom framework on top with explicit attribution. This keeps the planner interoperable while allowing organizations to add ISO, sector, or regional references without rewriting the core playbook.
Open-standards catalog
Use the links below as the reference catalog for open standards and governance resources. Do not reproduce normative text verbatim when the license posture does not allow it.
NIST AI RMF 1.0: https://doi.org/10.6028/NIST.AI.100-1
EUR-Lex EU AI Act: https://eur-lex.europa.eu/eli/reg/2024/1689
ISO/IEC 42001 (AI management systems): https://www.iso.org/standard/42001
ISO/IEC 23894 (AI risk management): https://www.iso.org/standard/77304.html
ISO/IEC 42005 (AI impact assessment): https://www.iso.org/standard/42005
Files
- ATTRIBUTION.mdAdded by AutoGineer956 B
- LICENSEAdded by AutoGineer16.6 KiB
- SKILL.md6.0 KiB
- references/ai-stride-overlay.md2.3 KiB
- references/eu-ai-act.md1.9 KiB
- references/nist-ai-rmf-govern.md2.3 KiB
- references/nist-ai-rmf-manage.md1.9 KiB
- references/nist-ai-rmf-map.md2.0 KiB
- references/nist-ai-rmf-measure.md2.1 KiB
references/ai-stride-overlay.md
title: AI STRIDE Overlay description: Threat-modeling overlay for RAI and security model analysis
AI STRIDE Overlay
This overlay extends STRIDE analysis for AI systems by capturing data, model, inference, feedback-loop, and human-review risks that standard software threat models often omit.
Core extensions
Training data stores
Model artifacts and checkpoints
Inference endpoints and serving layers
Feature pipelines and data preparation
Feedback loops and retraining paths
Human review queues and escalation paths
Monitoring dashboards and telemetry pipelines
Orchestration layers and agentic control flows
Trust-boundary guidance
Treat each training, inference, and monitoring boundary as a trust boundary. Evaluate whether data provenance, model versioning, telemetry, and human-review outputs can be altered, exfiltrated, or repudiated.
Dual threat-ID convention
T-RAI-### for RAI-managed threat IDs
T-{BUCKET}-AI-### for cross-references that connect to the Security Planner's bucketed analysis
ML STRIDE matrix
| STRIDE category | AI-specific concern |
|---|---|
| Spoofing | Identity confusion across model endpoints, service identities, and human-review actors |
| Tampering | Data poisoning, model tampering, feature drift injection |
| Repudiation | Missing audit trails for model decisions and review actions |
| Information Disclosure | Training data leakage, explanation leakage, inference exfiltration |
| Denial of Service | Resource exhaustion, adversarial examples, runaway retraining |
| Elevation of Privilege | Orchestrator or model-action escalation beyond permitted scope |
Merge protocol
When the Security Planner produces a security model, merge the RAI-specific findings into the existing plan using the dual threat-ID convention and preserve both the RAI-managed and cross-referenced IDs in the output.
Source attribution
This reference is adapted from the repository's RAI security-model guidance and is intended for reuse by the Security Planner and RAI Planner.
Requirements and compatibility
Written for: GitHub Copilot, Claude Code
Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta
“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.
Needs: Claude Code frontmatter
Security
No issues found by our static scan· No issues found
A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.
Attribution and licence
- Published by
- Microsoft
- Publisher
- Microsoft (Official)
- Source
- microsoft/hve-core · .github/skills/rai/rai-standards · commit f425e94
- Licence
- Creative Commons Attribution 4.0 International (redistribution allowed)
- LICENSE (written by AutoGineer)
- ATTRIBUTION.md (written by AutoGineer)
- Synced
AutoGineer is not affiliated with or endorsed by Microsoft.
Related skills
Gdpr Data Handling
Seth Hobson· Community
A practical guide for implementing GDPR-compliant data processing, consent management, and privacy controls.
Business and productivityClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository42.3k
Legal Advisor
Seth Hobson· Community
Drafts comprehensive, regulation-compliant legal documentation for businesses and software services.
Business and productivityClaude CodeMITGitHub stars of the source repository42.3k
Employment Contract Templates
Seth Hobson· Community
A library of templates and best practices for drafting professional employment contracts, offer letters, and HR policies.
Business and productivityClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository42.3k
Kyc Rules
AnthropicOfficial
Applies internal KYC and AML rules to onboarding records to determine risk ratings and document requirements.
Business and productivityClaude CodeClaude appsOpenAI Codex+14 moreApache-2.0GitHub stars of the source repository41.1k
Kyc Doc Parse
AnthropicOfficial
Extracts and structures data from client onboarding documents for KYC processing.
Business and productivityClaude CodeClaude appsOpenAI Codex+14 moreApache-2.0GitHub stars of the source repository41.1k
Signature Request
AnthropicOfficial
Streamline the preparation and routing of documents for electronic signature with a structured verification and configuration workflow.
Business and productivityClaude CodeClaude appsOpenAI Codex+14 moreApache-2.0GitHub stars of the source repository35.1k
Download ZIP
The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download rai-standards.zip