feat(mcsb): add Microsoft Cloud Security Benchmark security skill

.cspell.json

@@ -76,16 +76,19 @@
     "ASEC",
     "atheris",
     "ation",
+    "azuredeploy",
     "BABOK",
     "backlinked",
     "behaviour",
     "behavioural",
     "beval",
+    "blazor",
     "bmve",
     "boostsecurityio",
     "brainwriting",
     "calledprocesserror",
     "capabilit",
+    "chakra",
     "clarif",
     "clusterfuzzlite",
     "cmdletbinding",
@@ -108,6 +111,7 @@
     "fetchone",
     "figjam",
     "filenotfounderror",
+    "fluentui",
     "gcrm",
     "graphify",
     "graphifyy",
@@ -118,6 +122,7 @@
     "IIBA",
     "Infima",
     "ISTQB",
+    "langchain",
     "learning",
     "Linkified",
     "linkify",
@@ -143,7 +148,9 @@
     "poutine",
     "profanit",
     "prval",
+    "pubspec",
     "pylint",
+    "reakit",
     "refus",
     "revalidat",
     "scal",
@@ -161,13 +168,16 @@
     "subdirs",
     "terroris",
     "thiserror",
+    "treeviews",
     "trustworthi",
     "uncited",
     "underspecified",
+    "unremediated",
     "vally",
     "validat",
     "viab",
     "violen",
+    "vulnerab",
     "vulnerabilit",
     "vyta",
     "watchlist",

.github/agents/security/security-planner.agent.md

@@ -55,12 +55,13 @@ Each phase entry begins with a mandatory `read_file` of the indicated skill refe
 
 Beyond the always-load references above, load these specialized security skills only when the corresponding surface is present. Read them on entry to the phase noted, after the mandatory references. Skip any whose trigger is absent.
 
-| Trigger (from Phase 1 scoping / Phase 2 buckets) | Load on entry | Skill(s) to `read_file`                               |
-|--------------------------------------------------|---------------|-------------------------------------------------------|
-| AI/ML components detected (`raiEnabled` true)    | Phase 3 & 4   | `owasp-agentic`; `owasp-mcp` when MCP tooling is used |
-| `infrastructure` bucket present                  | Phase 3 & 4   | `owasp-infrastructure`                                |
-| `build` or `devops/platform-ops` bucket present  | Phase 3 & 4   | `owasp-cicd`, `supply-chain-security`                 |
-| Any project (cross-cutting GS overlay)           | Phase 4       | `secure-by-design`                                    |
+| Trigger (from Phase 1 scoping / Phase 2 buckets) | Load on entry | Skill(s) to `read_file`                                |
+|--------------------------------------------------|---------------|--------------------------------------------------------|
+| AI/ML components detected (`raiEnabled` true)    | Phase 3 & 4   | `owasp-agentic`; `owasp-mcp` when MCP tooling is used  |
+| `infrastructure` bucket present                  | Phase 3 & 4   | `owasp-infrastructure`                                 |
+| Azure cloud resources present                    | Phase 3 & 4   | `mcsb` (stable taxonomy; delegate per-service lookups) |
+| `build` or `devops/platform-ops` bucket present  | Phase 3 & 4   | `owasp-cicd`, `supply-chain-security`                  |
+| Any project (cross-cutting GS overlay)           | Phase 4       | `secure-by-design`                                     |
 
 If a conditional skill fails to load, note the gap and continue rather than halting. Activate `rpi-research` only for standards with no matching skill.
 
@@ -99,7 +100,7 @@ Gate: summary-and-advance — surface a brief phase summary and proceed unless t
 
 ### Phase 3: Standards Mapping
 
-Map controls from OWASP Top 10, NIST 800-53, and CIS Benchmarks to each bucket. Activate `rpi-research` for WAF and CAF lookups at runtime rather than embedding those standards directly.
+Map controls from OWASP Top 10, NIST 800-53, and CIS Benchmarks to each bucket. For Azure cloud resources, load the `mcsb` skill for the stable MCSB control-domain taxonomy and crosswalk, and delegate volatile per-service control lookups to the Researcher Subagent. Activate `rpi-research` for WAF and CAF lookups at runtime rather than embedding those standards directly.
 
 Human-review exit reminder: a qualified security reviewer confirms the standards-to-bucket mappings and any deferred lookups before advancing to Phase 4.
 

.github/agents/security/security-reviewer.agent.md

@@ -96,7 +96,7 @@ Report path pattern (plan): `.copilot-tracking/security/{{YYYY-MM-DD}}/plan-risk
 
 Sequence number resolution: Determine `{{NNN}}` by listing existing reports in the date directory, extracting the highest sequence number, incrementing by one, and zero-padding to three digits. Start at `001` when no reports exist.
 
-Skill resolution: Read the applicable security skill (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`) to access vulnerability references. Follow the skill's normative reference links to load vulnerability reference documents.
+Skill resolution: Read the applicable security skill (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`, `mcsb`) to access vulnerability references. Follow the skill's normative reference links to load vulnerability reference documents.
 
 ### Subagents
 
@@ -116,6 +116,7 @@ Skill resolution: Read the applicable security skill (e.g., `owasp-top-10`, `owa
 * owasp-infrastructure
 * owasp-cicd
 * secure-by-design
+* mcsb
 
 ## Subagent Prompt Templates
 

.github/agents/security/subagents/codebase-profiler.agent.md

@@ -34,7 +34,7 @@ Scan the repository to identify its technology stack and determine which securit
 
 ## Constants
 
-Skill resolution: Read the applicable security skill by name (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`). Resolve accessibility guidance through the consolidated Accessibility skill contract and use the matching framework guidance when needed.
+Skill resolution: Read the applicable security skill by name (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`, `mcsb`). Resolve accessibility guidance through the consolidated Accessibility skill contract and use the matching framework guidance when needed.
 
 ### Technology Signals
 
@@ -75,6 +75,12 @@ owasp-cicd:
   - "Jenkinsfile"
   - ".gitlab-ci.yml"
   - "azure-pipelines.yml"
+mcsb:
+  - "Azure Bicep files (.bicep)"
+  - "Azure Resource Manager (ARM) templates (azuredeploy.json)"
+  - "Terraform using the azurerm provider (.tf)"
+  - "azure-pipelines.yml targeting Azure resources"
+  - "Azure service references (Storage, Key Vault, AKS, SQL, App Service)"
 secure-by-design:
   - "SECURITY.md or security policy files"
   - "Threat model documents"

.github/agents/security/subagents/skill-assessor.agent.md

@@ -30,7 +30,7 @@ Assess exactly one security knowledge skill per invocation. Read all vulnerabili
 
 ## Constants
 
-Skill resolution: Read the applicable security skill by name (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`). Follow the skill's normative reference links to access the vulnerability index and individual vulnerability references.
+Skill resolution: Read the applicable security skill by name (e.g., `owasp-top-10`, `owasp-llm`, `owasp-agentic`, `owasp-mcp`, `owasp-infrastructure`, `owasp-cicd`, `secure-by-design`, `mcsb`). Follow the skill's normative reference links to access the vulnerability index and individual vulnerability references.
 
 ### Status Values
 
@@ -141,8 +141,8 @@ Make all guidance specific to the plan content rather than generic boilerplate.
 ### Step 1: Gather All Vulnerability References
 
 1. Read the located skill entry file and capture framework metadata (name, version, reference URL).
-2. Follow the entry file's normative reference links to read the vulnerability index (`references/00-vulnerability-index.md`) and extract the full list of vulnerability IDs.
-3. For each vulnerability ID in the index, read the corresponding reference file from the skill's `references/` directory and store its full content.
+2. Follow the entry file's normative reference links to read the skill's index file (for example `references/00-vulnerability-index.md` for vulnerability skills, `references/00-principle-index.md` for principle skills, or `references/00-control-index.md` for control or benchmark skills such as `mcsb`) and extract the full list of assessable items (vulnerability identifiers, principle identifiers, or control-domain identifiers). For benchmark skills, the control-domain identifiers in the index are the assessable grain; finer per-service control identifiers are volatile and retrieved separately per the skill's lookup playbook.
+3. For each item in the index, read the corresponding reference file from the skill's `references/` directory and store its full content.
 4. Do not proceed to Step 2 until every reference file has been read and stored.
 
 ### Step 2: Analyze Against References

.github/instructions/security/standards-mapping.instructions.md

@@ -18,7 +18,7 @@ The following standards are also delegated for runtime lookup due to version sen
 | Standard                                          | Rationale for Delegation                                   |
 |---------------------------------------------------|------------------------------------------------------------|
 | WAF / CAF                                         | Cloud-specific, frequently updated, extensive content      |
-| MCSB (Microsoft Cloud Security Benchmark)         | Azure-specific, frequently updated control mappings        |
+| MCSB per-service control mappings                 | Azure-specific, frequently updated per-service baselines   |
 | PCI-DSS                                           | Domain-specific, version-dependent compliance requirements |
 | S2C2F (Secure Supply Chain Consumption Framework) | Emerging standard, evolving maturity levels                |
 | SLSA (Supply Chain Levels for Software Artifacts) | Version-dependent build integrity requirements             |
@@ -35,10 +35,11 @@ When buckets or AI components from Phases 1–2 match, prefer the matching speci
 
 * AI/ML components → `owasp-agentic`, and `owasp-mcp` when MCP tooling is used (alongside the always-loaded `owasp-llm`)
 * `infrastructure` bucket → `owasp-infrastructure`
+* Azure cloud resources → `mcsb` (stable MCSB control-domain taxonomy and crosswalk; delegate volatile per-service control lookups)
 * `build` / `devops-platform-ops` buckets → `owasp-cicd`, `supply-chain-security`
 * Cross-cutting GS overlay → `secure-by-design`
 
-These skills are loaded by the Security Planner's Conditional Skill Map. Activate `rpi-research` only for standards with no matching skill (WAF, CAF, MCSB, PCI-DSS, and the others listed above).
+These skills are loaded by the Security Planner's Conditional Skill Map. Activate `rpi-research` only for standards with no matching skill (WAF, CAF, PCI-DSS, and the others listed above) and for the volatile per-service layer of MCSB.
 
 ### When to Activate Research
 

.github/skills/security/mcsb/SKILL.md

@@ -0,0 +1,52 @@
+---
+name: mcsb
+description: Microsoft Cloud Security Benchmark (MCSB v2) control-domain taxonomy and NIST 800-53 / CIS Controls crosswalk for planning and reviewing Azure cloud resources.
+license: MIT
+user-invocable: false
+metadata:
+  authors: "Microsoft"
+  spec_version: "1.0"
+  framework_revision: "MCSB v2 (preview)"
+  last_updated: "2026-07-21"
+  content_based_on: "https://learn.microsoft.com/security/benchmark/azure/overview"
+---
+
+# Microsoft Cloud Security Benchmark — Skill Entry
+
+This `SKILL.md` is the entrypoint for the Microsoft Cloud Security Benchmark (MCSB) skill.
+
+The skill encodes the durable, structurally stable layer of MCSB — the control-domain taxonomy and a domain-grain crosswalk to NIST SP 800-53 and CIS Controls — so the Security Planner and Security Reviewer can map and assess Azure cloud resources against a consistent control vocabulary.
+
+The skill deliberately does not embed the volatile layer of MCSB (per-Azure-service security baselines, per-service control IDs, Azure Policy mappings, and Defender for Cloud assessment specifics). That content changes on Microsoft's release cadence and is retrieved at runtime through the Researcher Subagent per [references/lookup-playbook.md](references/lookup-playbook.md).
+## Version and stability
+
+This skill targets **MCSB v2**, which Microsoft marks as preview and which supersedes MCSB v1. Content is version-pinned and retrieval-dated (2026-07-21). MCSB v2 replaces v1's Governance and Strategy (`GS`) domain with an Artificial Intelligence Security (`AI`) domain, and maps to NIST SP 800-53 Rev. 5 and CIS Controls v8.1 (v1 mapped to Rev. 4 and CIS v8). Re-verify the taxonomy and mappings against the official source before relying on them for a compliance decision.
+
+## Normative references
+
+1. [00 Control Index](references/00-control-index.md)
+2. [01 Network Security](references/01-network-security.md)
+3. [02 Identity Management](references/02-identity-management.md)
+4. [03 Privileged Access](references/03-privileged-access.md)
+5. [04 Data Protection](references/04-data-protection.md)
+6. [05 Asset Management](references/05-asset-management.md)
+7. [06 Logging and Threat Detection](references/06-logging-threat-detection.md)
+8. [07 Incident Response](references/07-incident-response.md)
+9. [08 Posture and Vulnerability Management](references/08-posture-vulnerability-management.md)
+10. [09 Endpoint Security](references/09-endpoint-security.md)
+11. [10 Backup and Recovery](references/10-backup-recovery.md)
+12. [11 DevOps Security](references/11-devops-security.md)
+13. [12 Artificial Intelligence Security](references/12-ai-security.md)
+14. [Lookup Playbook](references/lookup-playbook.md) — delegation guardrail for volatile per-service lookups.
+
+## Skill layout
+
+* `SKILL.md` — this file (skill entrypoint).
+* `references/` — the MCSB durable reference documents.
+  * `00-control-index.md` — control-domain catalog, consolidated crosswalk, and attribution.
+  * `01` through `12` — one document per MCSB v2 control domain with assessment checklists.
+  * `lookup-playbook.md` — delegation guardrail for volatile per-service content.
+
+## Attribution
+
+Reference content in this skill is original prose that paraphrases publicly documented MCSB structure. The MCSB v2 (preview) documentation this skill cites is published on Microsoft Learn from a non-public source repository and is governed by the Microsoft Learn Terms of Use, not a public Creative Commons license. The separate, older MicrosoftDocs/SecurityBenchmarks repository is CC BY 4.0, but its benchmark spreadsheets stop at Azure Security Benchmark v3 (the pre-rename lineage) and Microsoft Cloud Security Benchmark v1; it does not contain the MCSB v2 (preview) content this skill cites. Because this skill paraphrases rather than reproduces upstream text, no verbatim-reproduction license grant applies; it cites the canonical source in each reference file. See [references/00-control-index.md](references/00-control-index.md) for the consolidated attribution.