Contextualize workflow skills for Cybereum platform

[cybereum]

Summary

Updated all workflow skills (plan-ceo-review, plan-eng-review, ship, review, qa, retro) to be explicitly contextualized for the Cybereum capital project governance platform. This ensures that skill guidance, review criteria, and QA validation are aligned with Cybereum's domain (EVM, schedule intelligence, risk analysis, reference class forecasting) rather than generic web application patterns.

Key Changes

plan-ceo-review/SKILL.md:

• Added "Cybereum Product Context" section describing the platform, core components (temporal knowledge graph, Dyeus AI Engine, 8 analytical skills), target markets, and competitive positioning
• Updated philosophy and prime directives to emphasize capital project governance domain (e.g., "Platonic ideal of capital project governance AI" instead of generic "cathedral")
• Replaced generic engineering preferences with Cybereum-specific ones: calculation correctness against ANSI/EIA-748, temporal integrity, cross-skill consistency
• Refined scope expansion/hold/reduction modes with Cybereum context
• Simplified system audit commands and added skill-specific consistency checks

plan-eng-review/SKILL.md:

• Added "Cybereum Architecture Context" section mapping the 8 analytical skills, 6 workflow skills, and key data flows
• Updated Step 0 scope challenge to reference skills instead of files/classes
• Emphasized calculation correctness and cross-skill consistency in engineering preferences
• Clarified complexity thresholds (4 skills instead of 8 files)

qa/SKILL.md:

• Completely restructured from generic web app QA to Cybereum skill validation
• Replaced browser-based testing with calculation verification (EVM formulas, risk scoring, schedule health, completion prediction, reference class benchmarks)
• Added cross-skill consistency checks (terminology, JSON schemas, reference files)
• Updated issue taxonomy to focus on calculation errors, methodology compliance, and cross-skill coherence
• Removed browser automation setup; added skill inventory and calculation test cases

review/checklist.md:

• Replaced "SQL & Data Safety" with "Data & Calculation Integrity" (EVM formulas, risk scoring, schedule metrics, reference class data)
• Added "Graph & Data Consistency" checks for temporal knowledge graph mutations and schedule relationship logic
• Updated LLM output trust boundary to focus on AI-generated risk descriptions and Schwerpunkt recommendations
• Refined issue categories to match Cybereum domain (calculation correctness, cross-skill consistency, methodology compliance)

ship/SKILL.md:

• Added "for Cybereum capital project governance platform" to description
• Added build step to test/build workflow
• Removed Rails/JavaScript-specific test commands and eval suite logic
• Simplified to generic Cybereum test/build pattern

retro/SKILL.md:

• Added Cybereum context to description
• Added "Cybereum-aware: tracks skill development velocity and cross-skill consistency"

review/SKILL.md:

• Updated description to emphasize calculation integrity, graph consistency, and cross-skill coherence

CLAUDE.md:

• Updated header to "gstack development (Cybereum fork)"
• Added section explaining that analytical skills live in separate cybereum_Projects repo
• Clarified that this repo contains workflow skills only

Notable Implementation Details

• All changes are documentation/guidance updates; no code logic changes
• Maintained consistent terminology across all skills (e.g., "8 analytical skills", "temporal knowledge graph", "Dyeus AI Engine")
• Aligned severity definitions and issue categories across plan-ceo-review, plan-eng-review, qa, and review skills
• Preserved skill structure and workflow while recontextualizing examples and criteria for capital project domain
• QA skill transformation is most significant: replaced generic web app testing with domain-specific calculation validation and cross-skill consistency checks

https://claude.ai/code/session_01NbTWS982B1xJFJCyt6ZVig