juanmarchetto/agent-skills/full-eval

Full Evaluation: Strategic + Personal Assessment

Evaluate any idea, project, or life decision from every angle -- 6 business evaluators and 8 life advisors running in parallel, cross-referenced into a unified verdict, delivered as a professional PDF report.

One command. Complete assessment.

When to Use This

Use this when the decision spans both strategic viability (will it work?) and personal fit (should I do it, now?). For purely business analysis, use architect alone. For purely personal planning, use council alone.

Dependencies

This skill composes three existing systems:

The Flow

                         +-----------+
                         |   INPUT   |
                         | idea/path |
                         +-----+-----+
                               |
                    +----------+----------+
                    |                     |
              +-----v-----+        +-----v-----+
              |  ARCHITECT |        |  COUNCIL  |
              |  (parallel)|        | (parallel)|
              +-----+-----+        +-----+-----+
              |           |        |             |
        +-----+-----+    |  +-----+-----+      |
        | 6 business |    |  | 8 life    |      |
        | evaluators |    |  | advisors  |      |
        +-----+-----+    |  +-----+-----+      |
              |           |        |             |
        +-----v-----+    |  +-----v-----+      |
        | Architect  |    |  | Council   |      |
        | Synthesizer|    |  | Synthesizer|     |
        +-----+-----+    |  +-----+-----+      |
              |           |        |             |
              +-----+-----+-------+             |
                    |                            |
              +-----v-----------+               |
              | META-SYNTHESIZER |              |
              | Cross-reference  |              |
              | both outputs     |              |
              +---------+-------+              |
                        |                      |
                  +-----v-----+               |
                  |    PDF    |               |
                  |  REPORT   |               |
                  +-----------+               |

Step 1: Parse Input

Detect whether the input is:

• New idea: A description, concept, or question (e.g., "AI-powered meal planning app")
• Existing project: A path to a codebase (e.g., /home/user/my-saas-project)
• Life decision: A personal question with strategic dimensions (e.g., "Should I quit my job to build this?")

Step 2: Dispatch Architect (Strategic Evaluation)

Run the Architect system with all 6 evaluators in parallel:

1. Risk Assessor -- Always included. Identifies critical, significant, and monitoring-level risks.
2. Technical Architect -- Auto-included for software ideas. Evaluates feasibility, architecture, stack choices.
3. Market Analyst -- Auto-included for revenue-generating ideas. TAM/SAM/SOM, competitive landscape, monetization.
4. Impact Evaluator -- Social, educational, and community impact assessment.
5. Resource Strategist -- Cost-benefit analysis, timeline, team requirements, bootstrapping viability.
6. Innovation Scout -- Technology readiness, first-mover advantage, adjacent possible opportunities.

The Architect Synthesizer combines these into:

• Value Matrix: Commercial / Educational / Social scores (each 1-10)
• Opportunity Score: Overall 1-10 with 6 sub-scores
• Risk Assessment: GO / CONDITIONAL GO / NO-GO
• Architecture Proposal: For software ideas
• Prioritized Next Steps: Maximum 7 actions

Step 3: Dispatch Council (Personal Evaluation)

Run the Council system with all 8 advisors in parallel:

1. Life Coach -- Values alignment, life balance, holistic direction
2. Finance Advisor -- Budget impact, runway, cash flow, investment risk
3. Health Advisor -- Physical/mental health implications, burnout risk, sustainability
4. Career Advisor -- Career trajectory, skill development, positioning, network effects
5. Learning Advisor -- Skill gaps, learning curve, educational investment required
6. Systems Advisor -- Time management, workflow impact, automation opportunities
7. Relationships Advisor -- Social impact, family considerations, collaboration dynamics
8. Creative Advisor -- Creative fulfillment, innovation alignment, aesthetic satisfaction

The Council Synthesizer combines these into:

• Life Impact Assessment: How this decision affects all life domains
• Key Synergies: Actions that serve multiple personal domains
• Conflicts Resolved: Where advisors disagreed
• Prioritized Action Plan: Maximum 7 actions
• Watch List: Items to monitor

Step 4: Meta-Synthesis

The Meta-Synthesizer (see agents/meta-synthesizer.md) receives both outputs and:

1. Identifies alignment -- Where strategic opportunity and personal readiness both say "go" (strong signal)
2. Identifies conflicts -- Where the opportunity is great but personal circumstances say "wait" (or vice versa)
3. Produces conditions -- Specific requirements that must be met before proceeding
4. Generates a narrative -- Connects the strategic and personal dimensions into a coherent story
5. Issues a combined verdict -- GO / CONDITIONAL GO / NOT NOW / NO-GO with clear reasoning

Step 5: Generate PDF Report

Use the pdf-presentation skill to format the combined output as a professional document:

• Title page with the idea/project name and date
• Executive Summary (meta-synthesis verdict)
• Strategic Assessment section (Architect output: Value Matrix, Opportunity Score, Risk Assessment)
• Personal Assessment section (Council output: Life Impact, Action Plan)
• Cross-Reference Analysis (Meta-Synthesizer: alignments, conflicts, conditions)
• Combined Next Steps (merged and deduplicated from both systems)

Example Output

Given the input: "Should I quit my job to build an AI-powered meal planning app?"

Meta-Synthesis Verdict

VERDICT: CONDITIONAL GO
 
The opportunity scores 7.5/10 strategically -- strong market demand, feasible
technology, and good resource efficiency. However, the personal assessment
reveals critical constraints that require a phased approach.
 
CONFLICTS IDENTIFIED:
- Market Analyst says "move fast, the window is 12-18 months" but Finance
  Advisor flags only 3 months of personal runway
- Innovation Scout says "first-mover advantage is real" but Health Advisor
  flags existing burnout indicators from current role
- Career Advisor notes this aligns with long-term trajectory but Relationships
  Advisor flags the strain of a startup on family commitments
 
CONDITIONS FOR GO:
1. Validate the idea for 3 months while still employed (evenings/weekends MVP)
2. Secure 9 months of runway before transitioning (current savings + 6 months)
3. Hit 50 beta users and $500 MRR before giving notice
4. Establish a hard stop date: if conditions not met in 6 months, reassess
 
COMBINED SCORE: 6.8/10 (strategic 7.5 adjusted down by personal constraints)
 
NARRATIVE: The idea has genuine merit -- the meal planning space is growing,
AI personalization is a defensible differentiator, and your technical skills
are a strong fit. But your current personal situation (limited runway, burnout
risk, family commitments) means an immediate leap would be reckless. The smart
play is a structured transition: validate while employed, build runway, then
make the jump with evidence and a safety net.

Value Matrix (from Architect)

| Dimension              | Score | Key Driver                                    |
|------------------------|-------|-----------------------------------------------|
| Commercial             | 7/10  | $4B meal kit market, AI personalization moat   |
| Educational            | 6/10  | ML/nutrition knowledge transfer                |
| Social                 | 7/10  | Addresses food waste and dietary health        |
| Technical Feasibility  | 8/10  | Proven tech stack, APIs available               |
| Risk Level             | 6/10  | Competitive market, regulatory (nutrition)      |
| Resource Efficiency    | 8/10  | Solo MVP viable, low infrastructure cost        |

Life Impact Assessment (from Council)

| Domain         | Impact | Key Concern                                     |
|----------------|--------|-------------------------------------------------|
| Finance        | HIGH   | 3 months runway insufficient for startup         |
| Health         | HIGH   | Burnout risk -- current role already draining     |
| Career         | POSITIVE | Aligns with ML/product trajectory               |
| Relationships  | MODERATE | Startup hours strain family time                |
| Learning       | POSITIVE | Deep ML and nutrition domain knowledge           |
| Creative       | HIGH   | Strong creative fulfillment potential             |

Error Handling

The system is designed to be resilient when individual components encounter issues:

If Architect Fails Partially

• If 1-2 evaluators fail, the Architect Synthesizer proceeds with available evaluators and notes gaps in its output
• The Meta-Synthesizer marks affected dimensions as "insufficient data" and lowers confidence
• The PDF report includes a notice about which evaluators were unavailable

If Council Fails Partially

• If 1-2 advisors fail, the Council Synthesizer proceeds with available advisors and notes gaps
• The Meta-Synthesizer adjusts confidence and flags uncovered personal domains
• Critical failures (Life Coach or Finance) trigger a warning in the final verdict

If Architect Fails Completely

• The Meta-Synthesizer produces a personal-only assessment with a clear disclaimer
• The verdict is marked "INCOMPLETE -- strategic assessment unavailable"
• The PDF report omits the Strategic Assessment section

If Council Fails Completely

• The Meta-Synthesizer produces a strategic-only assessment with a clear disclaimer
• The verdict is marked "INCOMPLETE -- personal assessment unavailable"
• The PDF report omits the Personal Assessment section

If PDF Generation Fails

• The full evaluation results are displayed in the terminal as formatted text
• A markdown file is saved as a fallback

Language

All evaluators, advisors, synthesizers, and the meta-synthesizer respond in the same language the user writes in. The entire evaluation — including the PDF report — is produced in the user's language. Matched automatically without asking.

Structure

full-eval-skill/
├── SKILL.md                        # This file
├── agents/
│   └── meta-synthesizer.md         # Cross-references architect + council outputs
├── .claude-plugin/plugin.json      # Plugin metadata
├── README.md                       # Installation and usage guide
├── LICENSE                         # MIT License
└── .gitignore