Hybrid Decision Analysis

Produce a defensible decision by combining:

• skill-reviewer strengths: criteria confirmation, evidence notes, quality gates.
• comparative-analysis strengths: deterministic scoring and ranking.
• comparative-decision-review strengths: explicit score scale, effort and risk tie-breaks, coverage visibility.

Inputs

• Decision statement.
• Current platform.
• Primary user and workflow context.
• Hard constraints and non-goals.
• Time horizon.
• Initial alternatives or discovery scope.

Workflow

1. Frame decision and constraints.

2. Run discovery from references/discovery-protocol.md.

   • Include internal options, compose/extend variants, and external alternatives.
   • Include at least one external option unless explicitly blocked.

3. Normalize alternatives.

   • Require at least two alternatives before scoring.
   • Mark each alternative feasible or infeasible.

4. Derive criteria from intended use.

   • Use references/rubric-packs.md as a starting point, then adapt.

5. Confirm criteria, weights, scale, and platform set.

6. Prepare input using references/input-schema.md.

7. Score with guardrails:

   python3 skills/hybrid-decision-analysis/scripts/score_with_guardrails.py \
     --input <analysis-input.json> \
     --output <analysis-report.md> \
     --json-output <analysis-result.json>

8. Run reliability checks.

   • Use references/scenario-bakeoff-protocol.md for bakeoff evaluation.
   • Run sensitivity checks for close results.

9. Validate contracts:

   node skills/hybrid-decision-analysis/scripts/validate_json_contract.mjs \
     --schema skills/hybrid-decision-analysis/references/input.schema.json \
     --data <analysis-input.json>
   node skills/hybrid-decision-analysis/scripts/validate_json_contract.mjs \
     --schema skills/hybrid-decision-analysis/references/output.schema.json \
     --data <analysis-result.json>

10. Produce record using assets/hybrid-decision-record-template.md.

Required Guarantees

• Require at least two alternatives for recommendation scoring.
• Require criteria fields: metric, data_source, scoring_rule.
• Treat missing evidence as missing, not zero.
• Exclude missing criterion values from weighted averages.
• Track coverage and gate select or compose on minimum coverage.
• Never recommend an infeasible option unless all options are infeasible.
• Always expose decision status as proceed, defer, or no-go.
• Include explicit per-option justification in final output.
• Include all identified alternatives in the ranked record.
• Apply an agentic-eval loop (Generate → Evaluate → Critique → Refine), max 3 iterations.
• Use structured JSON for critique output and stop if no improvement between iterations.

Testing

Run guardrail tests before relying on a scoring-script change:

python3 skills/hybrid-decision-analysis/scripts/test_score_with_guardrails.py
node skills/hybrid-decision-analysis/scripts/validate_json_contract.mjs \
  --schema skills/hybrid-decision-analysis/references/input.schema.json \
  --data <analysis-input.json>
node skills/hybrid-decision-analysis/scripts/validate_json_contract.mjs \
  --schema skills/hybrid-decision-analysis/references/output.schema.json \
  --data <analysis-result.json>

References

• Discovery protocol: references/discovery-protocol.md
• Input contract: references/input-schema.md
• Input JSON Schema: references/input.schema.json
• Output contract: references/output-schema.md
• Output JSON Schema: references/output.schema.json
• Rubric packs: references/rubric-packs.md
• Quality gates: references/quality-gates.md
• Bakeoff protocol: references/scenario-bakeoff-protocol.md
• Reusable bakeoff fixture: assets/bakeoff-fixture.v1.json
• Reusable bakeoff results template: assets/bakeoff-results-template.v1.json
• Decision record template: assets/hybrid-decision-record-template.md