Governance
Decision Matrix
Posted Jun 8, 2026
Structured multi-criteria evaluation framework that scores alternatives, documents rationale, and produces auditable decision records.
Decision Matrix evaluates options against weighted criteria using quantitative and qualitative scoring. It generates structured decision frameworks, compares alternatives with transparent trade-off analysis, and exports auditable records for governance and compliance. Every evaluation is logged with evidence for regulatory review and internal audit.
What it does
6 actions, each one checked against your rules and recorded before it runs.
define_criteria
Establish weighted decision criteria and scoring scales for evaluation
score_alternatives
Rate each option against all defined criteria using consistent methodology
calculate_totals
Compute weighted scores and rank alternatives by aggregate performance
analyze_tradeoffs
Identify and document trade-offs between competing criteria and options
generate_rationale
Produce structured justification for rankings and recommendation
export_audit_record
Package decision framework, scores, and rationale into compliance-ready artifact
How it runs
Every run follows the same governed path, from request to recorded result.
- 1
User submits decision context, alternatives, and criteria with optional weights
- 2
Skill normalizes criteria into a structured evaluation framework with scoring scales
- 3
Each alternative is scored against every criterion using quantitative or qualitative assessment
- 4
Weighted totals are computed and alternatives are ranked from highest to lowest score
- 5
Trade-off analysis highlights where top alternatives excel or underperform
- 6
Complete decision matrix is exported with rationale, evidence pointers, and audit metadata
- 7
Results are logged and retained for 180 days to satisfy governance requirements
Governance & audit
What gets captured and verified on every single run.
Decision Evidence Capture
All evaluation steps, scores, and rationale are logged with immutable timestamps and operator attribution
Criteria Transparency
Weights, scoring scales, and assessment methodology are recorded in the audit trail to prevent post-hoc manipulation
Data Retention Enforcement
Decision records are retained for 180 days in compliance with governance policy and purged automatically thereafter
Access Control Logging
Every read and export of decision artifacts is logged with user identity and timestamp for compliance review
Example in action
Vendor Selection for Enterprise Cloud Migration
- 1
Define Evaluation Criteria. Procurement team establishes criteria: cost (30% weight), compliance certifications (25%), technical capabilities (25%), support SLA (20%)
- 2
Score Three Vendor Proposals. Skill assesses each vendor against all criteria using proposal documents, pricing sheets, and compliance matrices
- 3
Calculate Weighted Rankings. Vendor A scores 82/100, Vendor B scores 78/100, Vendor C scores 71/100 after applying weights
- 4
Analyze Trade-offs. Vendor A leads on compliance and support but is 15% more expensive; Vendor B offers better cost-performance ratio but weaker certifications
- 5
Export Auditable Record. Complete decision matrix with scores, rationale, and evidence is exported for CFO and compliance review, then logged for audit trail
Skill manifest
The exact instructions this skill ships with, shown in full. Nothing hidden.
---
id: decision-matrix
version: 0.1.0
display_name: Decision Matrix
pack: governance
category: compliance
risk_tier: medium
tags: ['decision-support', 'criteria-analysis', 'governance', 'evaluation', 'audit-trail', 'risk-assessment']
input_schema:
type: object
required: [decision_context, alternatives, criteria]
properties:
decision_context:
type: string
description: Brief description of the decision to be made and its business context.
alternatives:
type: array
description: List of options or alternatives to evaluate.
items:
type: object
properties:
name:
type: string
description:
type: string
criteria:
type: array
description: Evaluation criteria with optional weights (defaults to equal if not provided).
items:
type: object
properties:
name:
type: string
description:
type: string
weight:
type: number
scale_min:
type: number
scale_max:
type: number
supporting_documents:
type: array
description: Optional references or documents to inform scoring.
items:
type: string
output_schema:
type: object
required: [ranked_alternatives, decision_matrix, rationale, audit_record]
properties:
ranked_alternatives:
type: array
description: Alternatives ordered by weighted score from highest to lowest.
items:
type: object
properties:
name:
type: string
total_score:
type: number
rank:
type: integer
decision_matrix:
type: object
description: Structured table of scores for each alternative against each criterion.
rationale:
type: string
description: Narrative justification for the ranking and key trade-offs identified.
tradeoff_analysis:
type: array
description: Specific trade-offs between top-ranked alternatives.
items:
type: object
audit_record:
type: object
description: Compliance-ready artifact with timestamps, evidence pointers, and methodology.
planner:
mode: dynamic
prompt_section: "How to think about this task"
max_steps: 15
max_duration_sec: 300
max_cost_usd: 0.2
model_tier: balanced
allowed_step_kinds: [tool_call, llm_call, output]
evidence:
required_types: ['log']
capture_per_step: true
retention_days: 180
routing:
intent_examples:
- "Compare three vendors for our cloud migration project"
- "Build a decision matrix for product roadmap prioritization"
- "Evaluate hiring candidates using weighted criteria"
- "Score investment opportunities against risk and return factors"
not_for:
- "Make the final decision for me without showing the framework"
- "Run a single yes/no approval"
verifier_rules:
- "Every alternative must be scored against every criterion"
- "Weighted totals must sum correctly and alternatives must be ranked by descending score"
- "Audit record must contain methodology, timestamps, and evidence pointers"
---
## Overview
Decision Matrix provides a structured, multi-criteria evaluation framework for comparing alternatives in governance-sensitive contexts. It scores options against weighted criteria, computes rankings, and generates auditable decision records with full rationale. The skill is designed for procurement, vendor selection, investment prioritization, and risk-assessment scenarios where transparency and traceability are mandatory.
## How to think about this task
Your goal is to transform an unstructured decision problem into a rigorous, defensible evaluation. Start by normalizing the user's criteria into a consistent framework with explicit weights and scoring scales. If weights are missing, default to equal weighting and note that in the output.
Score each alternative against every criterion independently. Use quantitative data where available; for qualitative factors, apply a consistent rubric based on the scale provided. Document the rationale for each score, especially where judgment calls are required.
Once all scores are recorded, compute weighted totals and rank alternatives. Identify meaningful trade-offs between the top contenders: where does the highest-ranked option underperform relative to others, and what criteria drive the separation? Capture this analysis in plain language.
Finally, package the entire evaluation into an audit-ready artifact. Include methodology, raw scores, weights, timestamps, and references to any supporting documents. This record must be exportable and defensible in front of auditors, compliance officers, or executive committees.
## Capabilities
- Establish weighted decision criteria and scoring scales for evaluation
- Rate each option against all defined criteria using consistent methodology
- Compute weighted scores and rank alternatives by aggregate performance
- Identify and document trade-offs between competing criteria and options
- Produce structured justification for rankings and recommendation
- Package decision framework, scores, and rationale into compliance-ready artifact
## Inputs
**decision_context**: A brief description of the decision to be made, including business context, stakeholders, and any regulatory or policy constraints that govern the evaluation.
**alternatives**: An array of options to evaluate. Each alternative should include a name and description. The skill will score each one independently against the criteria.
**criteria**: The evaluation dimensions, each with a name, description, optional weight (defaults to equal if omitted), and optional scoring scale (defaults to 0–10). Weights should sum to 100 or 1.0 depending on convention.
**supporting_documents** (optional): References to documents, data sources, or prior analyses that inform the scoring process. The skill may retrieve or cite these during evaluation.
## Outputs
**ranked_alternatives**: A sorted list of alternatives with their weighted total scores and rank positions. The highest-scoring alternative appears first.
**decision_matrix**: A structured table showing the raw score for each alternative on each criterion, plus weighted contributions and totals.
**rationale**: A narrative explanation of the ranking, highlighting which criteria drove the outcome and where subjective judgment was applied.
**tradeoff_analysis**: A breakdown of how the top alternatives compare on individual criteria, surfacing where lower-ranked options may outperform the winner.
**audit_record**: A compliance-ready artifact containing methodology, timestamps, operator attribution, evidence pointers, and retention metadata. This record is logged and retained for 180 days.
## Tools
- **connectors**: none (operates on user-supplied data and documents)
- **operators**: scoring, aggregation, ranking, narrative generation
- **primitives**: structured output formatting, evidence logging, audit packaging
## Guardrails
- All evaluation steps, scores, and rationale are logged with immutable timestamps and operator attribution
- Weights, scoring scales, and assessment methodology are recorded in the audit trail to prevent post-hoc manipulation
- Decision records are retained for 180 days in compliance with governance policy and purged automatically thereafter
- Every read and export of decision artifacts is logged with user identity and timestamp for compliance review
## Examples
**Scenario**: Vendor Selection for Enterprise Cloud Migration
A procurement team needs to select a cloud infrastructure vendor for a multi-year migration. They define four criteria: cost (30% weight), compliance certifications (25%), technical capabilities (25%), and support SLA (20%). Three vendors submit proposals.
The team invokes Decision Matrix with:
- **decision_context**: "Select cloud vendor for 3-year enterprise migration supporting 50 applications and 2,000 users, governed by SOC 2 and GDPR requirements"
- **alternatives**: Vendor A, Vendor B, Vendor C (each with proposal summary)
- **criteria**: cost, compliance, technical capabilities, support SLA (with weights as above)
The skill scores each vendor:
- Vendor A: cost 7/10, compliance 10/10, technical 9/10, support 9/10 → weighted total 8.65
- Vendor B: cost 9/10, compliance 7/10, technical 8/10, support 8/10 → weighted total 8.00
- Vendor C: cost 8/10, compliance 6/10, technical 7/10, support 7/10 → weighted total 7.05
The output ranks Vendor A first but highlights a trade-off: Vendor A is 15% more expensive than Vendor B but leads on compliance and support. The rationale notes that the compliance advantage justifies the cost premium given regulatory requirements.
The audit record is exported with all scores, weights, methodology, and references to the three proposals. The procurement VP reviews and approves the recommendation. The record is logged for 180 days to satisfy internal audit and external SOC 2 controls.Reviews
Reviews from teams using Decision Matrix.
No reviews yet
Be the first to share how Decision Matrix works for your team.
Discussion(0)
No comments yet. Be the first.