Skill v1.0.0

currentAutomated scan100/100

security-phoenix-demo/security-skills-claude-code/phoenix-scope-cutter

──Details

PublishedSeptember 30, 2026 at 08:20 AM

Content Hashsha256:fc75b85e1751cbd2...

Git SHA40ace4aa7101

──Files

Files (1 file, 3.3 KB)

SKILL.md3.3 KBactive

SKILL.md · 105 lines · 3.3 KB

version: "1.0.0" name: phoenix-scope-cutter description: > Role 02 of the Phoenix Security spec pipeline. Converts CLEAN_CONTEXT into an explicit, minimal SCOPE_DEFINITION — goals, non-goals, in/out boundaries, assumptions, success metrics, and gating questions — anchored to Phoenix's product pillars and 12-person team constraints. Use this skill when you have a cleaned context block and need to define what the feature actually is and isn't, prevent scope spiral, or when someone says "define the scope", "what are we building vs not building", "set the boundaries for this feature", or "trim the scope". Always run after phoenix-context-curator and before phoenix-constraint-distiller.

Phoenix Security — Scope Cutter (Role 02)

Token Budget: ≤800 tokens Depends On: Role 01 — CLEAN_CONTEXT Feeds Into: Role 03 — Constraint Distiller Output: 02-scope-definition.md

Phoenix Scope Anchors

Every goal MUST map to at least one Phoenix product pillar:

Pillar	Scope signals
ASPM	Posture scoring, risk aggregation, multi-tool ingestion
CTEM	Exposure tracking, prioritisation loops, remediation workflows
Container Lineage	Image → deployment → vuln tracing
Reachability	Code-path analysis, false-positive reduction
Supply Chain / SCA	SBOM, dependency graph, Snyk integration
DevSecOps	CI/CD gating, shift-left, Jenkins / GitHub Actions
LLM Security	AI model risk, prompt injection surface
Board-Level Reporting	10-category risk, exec dashboards, compliance mapping

Team Constraint

12-person team. Any scope implying >2 sprint capacity without P0 justification → flag for @Alfonso Eusebio.

Success Metric Benchmarks

Vuln reduction: ClearBank 98%, Bazaarvoice 94%
False positive reduction: IAS 78%
Cost savings: ClearBank $15M, Bazaarvoice $6.3M, IAS $1.7M+
Rule auto-mapping: Bazaarvoice 32K via Backstage

Core Rules

Do not invent features, integrations, policies, or architecture.
Every NON_GOAL must be explicit — silence is not a boundary.
Flag scope touching active enterprise accounts (Mimecast, JM, Anaplan) → P0 by default.
Flag anything implying competitive positioning vs Prisma/Wiz/Snyk → route to @Francesco Cipollone.
If GOALS cannot be stated from input, output only SCOPE_OPEN_QUESTIONS.

Output Schema

markdown

---
meta:
  role: 02-scope-cutter
  timestamp: [ISO 8601]
  session_id: session-[YYYYMMDD-HHMMSS]
  token_count: [actual]
  status: complete | blocked | needs-input
  validation: passed | failed
  next_role: 03-constraint-distiller
---
 
### SCOPE_DEFINITION
 
#### GOALS (max 6)
-G1: ... (pillar: ASPM|CTEM|Container Lineage|Reachability|SCA|DevSecOps|LLM|Reporting)
 
#### NON_GOALS (max 6)
-NG1: ...
 
#### IN_SCOPE (max 10)
-IS1: ...
 
#### OUT_OF_SCOPE (max 10)
-OS1: ...
 
#### ASSUMPTIONS (max 6; map to FACT/DECISION IDs)
-A1: ... (based_on: F? / D?)
 
#### SUCCESS_METRICS (max 5; quantified where possible)
-SM1: ... (benchmark: ClearBank|Bazaarvoice|IAS|new)
 
#### SCOPE_OPEN_QUESTIONS (max 7; blockers only)
-SQ1: ...
 
#### ENTERPRISE_FLAGS (only if active accounts affected)
-EF1: <account> → <impact> → <escalation: @Francesco|@Alfonso|@Philip>

Chaining

Save: outputs/{session-id}/02-scope-definition.md → Pass to: phoenix-constraint-distiller (Role 03)

All versions