How do I install the Multi Perspective Analysis skill?

Run npx get-skill add tomevault-io/multi-perspective-analysis in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Multi Perspective Analysis safe to use?

SkillMD's automated safety review verdict for this skill is PASS. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Multi Perspective Analysis?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Multi Perspective Analysis free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Multi Perspective Analysis?

tomevault-io (@tomevault-io) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Multi Perspective Analysis by tomevault-io

Name: Multi Perspective Analysis
Author: tomevault-io

File contents

Multi-Perspective Analysis

Use the council_query tool to submit complex questions for adversarial deliberation by five personas, each representing a distinct intellectual tradition. Use council_decision when the user needs a concrete recommendation between options.

When to Use

Complex questions with no clear right answer
High-stakes decisions where multiple viewpoints matter
Research evaluation to stress-test findings
Strategic planning to surface blind spots
Contentious topics where bias is likely

The Five Personas

Persona	Tradition	Role
Advocate	Steel-manning	Builds the strongest case FOR the position
Skeptic	Popperian falsificationism	Finds the observation that kills the claim
Oracle	Empirical base-rate reasoning	Grounds debate in historical data and statistics
Contrarian	Kuhnian paradigm critique	Rejects the framing, proposes alternative paradigms
Arbiter	Bayesian synthesis	Synthesizes all views with explicit prior/posterior updates

How to Use

Basic Query

council_query(question="Should we migrate our monolith to microservices?")

With Context

council_query(
    question="Is proof-of-stake more secure than proof-of-work?",
    context="We're evaluating consensus mechanisms for a new L1 blockchain targeting institutional users."
)

Subset of Personas

council_query(
    question="Will quantum computing break RSA by 2030?",
    personas=["skeptic", "oracle", "arbiter"]
)

Cost Modes

council_query(question="Should we rewrite the storage layer?", mode="fast")
council_query(question="Should we rewrite the storage layer?", mode="standard")
council_query(question="Should we rewrite the storage layer?", mode="deep")

fast: Skeptic + Arbiter for cheap pre-checks
standard: Full council for normal decisions
deep: Full council plus a second Arbiter pass for high-stakes decisions

Decision Review

council_decision(
    options=["Ship now", "Delay for audit", "Cut scope and ship"],
    decision_context="Hackathon demo is tomorrow; auth changes are unreviewed",
    criteria=["demo impact", "safety", "reversibility"]
)

Interpreting Results

The council returns a structured verdict:

confidence_score (0-100): Arbiter's posterior confidence after weighing all arguments
conflict_detected: True if personas strongly disagree (confidence spread > 30%)
persona_responses: Each persona's analysis with their individual confidence
arbiter_synthesis: The Arbiter's Bayesian synthesis with prior/posterior reasoning
action_summary: Recommendation, top risks, missing evidence, and next actions
dpo_pairs: Preference pairs for training (aligned vs overruled responses)
sources: Model-cited URLs
verified_sources: URLs actually retrieved by the council evidence layer

Confidence Guidelines

80-100: Strong consensus, high-quality evidence
60-79: Moderate confidence, some unresolved tensions
40-59: Mixed signals, significant uncertainty
0-39: Low confidence, major disagreements or insufficient evidence

Workflow: Research Then Evaluate

For best results, combine web research with council evaluation:

Use web_search to gather initial data
Use web_extract to read key sources
Synthesize findings into a draft analysis
Use council_evaluate to stress-test your analysis
Revise based on council feedback

Custom Personas

Add custom personas in ~/.hermes-council/config.yaml:

personas:
  security_analyst:
    tradition: "Adversarial security thinking"
    system_prompt: "You are a security analyst. Find every attack vector..."
    tags: ["security", "adversarial"]

Custom personas are merged with defaults. Use the same name to override a default.

Source: Ridwannurudeen/hermes-council — distributed by TomeVault.

---name: multi-perspective-analysisdescription: Run adversarial multi-perspective analysis on complex questions using the council tool. Five personas debate from distinct intellectual traditions. Use when this capability is needed.---# Multi-Perspective AnalysisUse the `council_query` tool to submit complex questions for adversarial deliberation by five personas, each representing a distinct intellectual tradition. Use `council_decision` when the user needs a concrete recommendation between options.## When to Use- **Complex questions** with no clear right answer- **High-stakes decisions** where multiple viewpoints matter- **Research evaluation** to stress-test findings- **Strategic planning** to surface blind spots- **Contentious topics** where bias is likely## The Five Personas| Persona | Tradition | Role ||---------|-----------|------|| **Advocate** | Steel-manning | Builds the strongest case FOR the position || **Skeptic** | Popperian falsificationism | Finds the observation that kills the claim || **Oracle** | Empirical base-rate reasoning | Grounds debate in historical data and statistics || **Contrarian** | Kuhnian paradigm critique | Rejects the framing, proposes alternative paradigms || **Arbiter** | Bayesian synthesis | Synthesizes all views with explicit prior/posterior updates |## How to Use### Basic Query```council_query(question="Should we migrate our monolith to microservices?")```### With Context```council_query( question="Is proof-of-stake more secure than proof-of-work?", context="We're evaluating consensus mechanisms for a new L1 blockchain targeting institutional users.")```### Subset of Personas```council_query( question="Will quantum computing break RSA by 2030?", personas=["skeptic", "oracle", "arbiter"])```### Cost Modes```council_query(question="Should we rewrite the storage layer?", mode="fast")council_query(question="Should we rewrite the storage layer?", mode="standard")council_query(question="Should we rewrite the storage layer?", mode="deep")```- `fast`: Skeptic + Arbiter for cheap pre-checks- `standard`: Full council for normal decisions- `deep`: Full council plus a second Arbiter pass for high-stakes decisions### Decision Review```council_decision( options=["Ship now", "Delay for audit", "Cut scope and ship"], decision_context="Hackathon demo is tomorrow; auth changes are unreviewed", criteria=["demo impact", "safety", "reversibility"])```## Interpreting ResultsThe council returns a structured verdict:- **confidence_score** (0-100): Arbiter's posterior confidence after weighing all arguments- **conflict_detected**: True if personas strongly disagree (confidence spread > 30%)- **persona_responses**: Each persona's analysis with their individual confidence- **arbiter_synthesis**: The Arbiter's Bayesian synthesis with prior/posterior reasoning- **action_summary**: Recommendation, top risks, missing evidence, and next actions- **dpo_pairs**: Preference pairs for training (aligned vs overruled responses)- **sources**: Model-cited URLs- **verified_sources**: URLs actually retrieved by the council evidence layer### Confidence Guidelines- **80-100**: Strong consensus, high-quality evidence- **60-79**: Moderate confidence, some unresolved tensions- **40-59**: Mixed signals, significant uncertainty- **0-39**: Low confidence, major disagreements or insufficient evidence## Workflow: Research Then EvaluateFor best results, combine web research with council evaluation:1. Use `web_search` to gather initial data2. Use `web_extract` to read key sources3. Synthesize findings into a draft analysis4. Use `council_evaluate` to stress-test your analysis5. Revise based on council feedback## Custom PersonasAdd custom personas in `~/.hermes-council/config.yaml`:```yamlpersonas: security_analyst: tradition: "Adversarial security thinking" system_prompt: "You are a security analyst. Find every attack vector..." tags: ["security", "adversarial"]```Custom personas are merged with defaults. Use the same name to override a default.---> Source: [Ridwannurudeen/hermes-council](https://github.com/Ridwannurudeen/hermes-council) — distributed by [TomeVault](https://tomevault.io).

Frequently asked questions

Multi Perspective Analysis