Generated banner for Security Reviewer Formats
Generated logo for Security Reviewer Formats

Security Reviewer Formats

MicrosoftOfficialPublished by Microsoft

  • Skill
  • Guidelines
  • Software development
  • MIT
  • No issues found

From the author

Format specifications and data contracts for the security reviewer orchestrator and its subagents.
— the description by Microsoft

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

This skill provides a centralized repository of format specifications and data contracts designed for security review orchestrators and their subagents. It ensures consistency across the reporting pipeline by defining standardized structures for vulnerability findings, scan status updates, completion summaries, and severity definitions. By establishing these shared schemas, the skill enables reliable data exchange between different components of the assessment process, including audit, diff, and plan modes.

What it does

  • Defines standardized templates for vulnerability, plan, and RAI reports.
  • Specifies serialization formats for security and RAI findings.
  • Provides consistent status update and completion report structures.
  • Establishes uniform severity level definitions for OWASP assessments.
  • Defines data contracts for inter-subagent communication and verification results.

When to use it

  • When developing or configuring subagents for a security review orchestration system.
  • When implementing new reporting modules that must integrate with existing assessment workflows.
  • When standardizing the output format of security scans across different assessment modes.
  • When ensuring consistent severity and status reporting across multiple security skills.

Advantages

  • Ensures interoperability between different subagents and the main orchestrator.
  • Provides a single source of truth for reporting and finding schemas.
  • Facilitates consistent data exchange and verification processes.
  • Simplifies the integration of new security assessment skills into the existing framework.

Usage examples

“What is the required format for a vulnerability finding serialization?”
The agent provides the specific markdown structure including ID, Title, Status, Severity, Location, Finding, and Recommendation fields.
“How should a scan completion report be structured for an audit mode assessment?”
The agent returns the template containing the report path, mode, skills assessed, severity counts, verification statistics, and summary counts.

Install

  1. Download the skill as a ZIP:

    https://mp-staging.autogineer.com/agent-skills/microsoft/security-reviewer-formats/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip security-reviewer-formats.zip -d .agents/skills

Documentation by Microsoft

Security Reviewer Formats — Skill Entry

This SKILL.md is the entrypoint for the security reviewer format specifications skill.

The skill provides shared format templates and data contracts used by security, accessibility, and RAI review orchestrators and their subagents. Each reference file covers a focused area of the reporting pipeline.

Normative references

  1. Report Formats — VULN_REPORT_V1, PLAN_REPORT_V1, and RAI_REPORT_V1 templates.

  2. Finding Formats — Security and RAI finding, verification, and collection formats.

  3. Completion Formats — Scan status, completion, RAI response, and terminal error formats.

  4. Severity Definitions — Standard severity level definitions for all OWASP skill assessments.

Skill layout

  • SKILL.md — this file (skill entrypoint).

  • references/ — format specification documents.

    • report-formats.md — full report templates for audit, diff, and plan modes.

    • finding-formats.md — serialization and collection formats for findings exchange between subagents.

    • completion-formats.md — status updates, completion summaries, and the minimal profile stub.

    • severity-definitions.md — severity level table shared across all assessments.

Files

references/finding-formats.md


title: Finding Formats description: Finding Serialization Format and Verified Findings Collection Format for inter-subagent data exchange

Finding Formats

Finding Serialization Format

Each finding passed to the Finding Deep Verifier is a markdown block with these fields:

- **ID:** <FINDING_ID>
- **Title:** <FINDING_TITLE>
- **Status:** <FINDING_STATUS>
- **Severity:** <FINDING_SEVERITY>
- **Location:** <FILE_LOCATION>
- **Finding:** <FINDING_DESCRIPTION>
- **Recommendation:** <RECOMMENDATION>

Verified Findings Collection Format

The merged collection of verified findings passed to Report Generator uses the following structure:

  • Items are grouped by skill name.

  • UNCHANGED items (PASS and NOT_ASSESSED) use the Finding Serialization Format with an added - **Verdict:** UNCHANGED field.

  • Verified items (FAIL and PARTIAL after deep verification) include the full Deep Verification Verdict block as returned by Finding Deep Verifier.

### owasp-top-10

## Finding: A01-001: Broken Access Control, Missing authorization checks

### Original Assessment
- **Status:** FAIL
- **Severity:** High
- **Finding:** No authorization middleware on admin endpoints.

### Vulnerability Reference Analysis
- **Reference file:** .github/skills/security/owasp-top-10/references/A01-broken-access-control.md
- **Applicable checklist items:** A01-001, A01-003
- **Attack preconditions:** Unauthenticated network access to the API

### Vulnerable Location
- **File:** [src/api/routes.ts#L45](src/api/routes.ts#L45)
- **Lines:** L42-L48

### Offending Code

```ts
app.get('/api/admin/users', (req, res) => {
  return db.users.findAll();
});

RAI Finding and Verification Formats

RAI_FINDING_V1 contains ID, Title, Status, Severity, Location, Finding, and Recommendation. For verifier input, Status is FAIL or PARTIAL and every field is required.

RAI_DEEP_VERIFICATION_V1 contains Schema, Finding ID, Finding Title, Original Status, Original Severity, Framework, Requirement, Confirming Evidence, Contradicting Evidence, Necessary Capability Assessment and rationale, Verdict, Verified Status, Verified Severity, Verified Categories, Justification, Updated Recommendation, and Limitations. Required verdicts are CONFIRMED, DISPROVED, and DOWNGRADED. Necessary Capability Assessment is REQUIRED, AVOIDABLE, or NOT_APPLICABLE. Verified Categories contains evidence-supported AI STRIDE categories, NO_CATEGORY, or NOT_ASSESSED.

The total mapping is: CONFIRMED retains the source-supported FAIL or PARTIAL status, severity, and residual categories; DOWNGRADED maps to PARTIAL with lower evidence-supported severity; and DISPROVED maps to PASS, NONE, and NO_CATEGORY. A verifier never emits UNCHANGED.

RAI_VERIFIED_FINDINGS_COLLECTION_V1 groups each framework's findings. FAIL and PARTIAL items are exactly one RAI_DEEP_VERIFICATION_V1 response; PASS and NOT_ASSESSED items are RAI_FINDING_V1 records with Verdict=UNCHANGED. Every input finding occurs exactly once.

RAI_PLAN_FINDINGS_COLLECTION_V1 groups framework findings with ID, Title, Status, Severity, Finding, Recommendation, and optional limitation. Status is RISK, CAUTION, COVERED, or NOT_APPLICABLE. Verification fields are prohibited.

Confirming Evidence
  • No auth middleware registered on the admin router at src/api/routes.ts#L42-L48.

  • No global middleware guard for /api/admin/* routes.

Contradicting Evidence

None found.

Verdict
  • Verdict: CONFIRMED

  • Verified Status: FAIL

  • Verified Severity: High

  • Justification: Endpoint /api/admin/users lacks any auth middleware. Direct access confirmed via route definition at src/api/routes.ts:45. No global or route-level guards exist.

Updated Remediation

Add authorization middleware that checks for admin role before processing admin API requests.

Example Fix
app.get('/api/admin/users', requireRole('admin'), (req, res) => {
  return db.users.findAll();
});

  • ID: A01-002

  • Title: Broken Access Control: CORS misconfiguration

  • Status: PASS

  • Severity: N/A

  • Location: N/A

  • Finding: CORS configuration restricts origins appropriately.

  • Recommendation: N/A

  • Verdict: UNCHANGED

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/security/security-reviewer-formats · commit f425e94
Licence
MIT License (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by Microsoft.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download security-reviewer-formats.zip