Skills Overview

Skills are the WHAT of AI Mantras - actionable capabilities that personas can invoke to accomplish specific types of work. During orchestration, the orchestrator selects appropriate skills for each persona based on the task requirements.

Understanding Skills

Skills complement the other pillars:

Pillar	Question	Purpose
Personas	WHO	Define identity, expertise, style
Patterns	HOW	Define reasoning structures
Principles	WHY	Define ethics and values
Skills	WHAT	Define tools and actions

Skill Categories

Research Skills Information gathering: web search, document review, codebase exploration, literature review.

Analysis Skills Analytical capabilities: data analysis, code review, financial modeling, risk assessment.

Creation Skills Content generation: document generation, code generation, diagrams, plans, reports.

Evaluation Skills Assessment capabilities: quality check, security audit, compliance review, fact verification.

Orchestration Skills Workflow coordination: task decomposition, handoff, summarization, progress tracking.

Utility Skills Framework navigation: help and discovery.

Skill-Persona Affinity

Each persona has primary and secondary skills that match their expertise:

Persona	Primary Skills	Secondary Skills
Clara (Finance)	financial-modeling, risk-assessment, data-analysis	comparative-analysis, report-writing
Kestra (Systems)	codebase-exploration, security-audit, code-review	diagram-creation, risk-assessment
Watson (Medical)	literature-review, fact-verification, risk-assessment	document-review, report-writing
Goeth (Philosophy)	comparative-analysis, document-review	summarization, report-writing
Franklin (Reasoning)	comparative-analysis, risk-assessment, data-analysis	literature-review, fact-verification
Ada (QA)	quality-check, fact-verification	code-review, compliance-review
Drucker (Goals)	quality-check, progress-tracking	comparative-analysis
Rickover (Safety)	security-audit, risk-assessment, compliance-review	fact-verification
Bernstein (Orchestrator)	task-decomposition, handoff, constraint-gathering	progress-tracking, summarization
Hopper (Planner)	task-decomposition, plan-drafting, constraint-gathering	risk-assessment
Lovell (Crisis)	task-decomposition, risk-assessment, progress-tracking	handoff, constraint-gathering

How Skills Are Used

During Planning Phase

The orchestrator (Bernstein/Hopper) reviews the task and selects skills:

For this infrastructure audit task:
- Kestra will use: codebase-exploration, security-audit, risk-assessment
- Patterns: threat-modeling, chain-of-thought
- Deliverable: Security assessment report

During Execution Phase

Domain personas invoke skills as needed:

[Kestra applying security-audit skill]
Following security-audit.md protocol:
1. Identify attack surface...
2. Enumerate vulnerabilities...
3. Assess impact and likelihood...

During Evaluation Phase

Evaluators use evaluation skills:

[Ada applying quality-check skill]
Reviewing against success criteria using quality-check.md:
- Criterion 1: Coverage - PASS
- Criterion 2: Accuracy - NEEDS REVISION