necessity-evaluation

$npx mdskill add yogsoth-ai/de-anthropocentric-research-engine/necessity-evaluation

Evaluates probability of necessity for a causal factor

  • Determines if a conclusion fails when a factor is absent
  • Uses counterfactual reasoning via subagent execution
  • Analyzes artifact, factor, and conclusion inputs for necessity
  • Returns PN score, reasoning, alternatives, and confidence

SKILL.md

.github/skills/necessity-evaluationView on GitHub ↗
---
name: necessity-evaluation
description: Evaluate the probability of necessity (PN) for a causal factor — would the conclusion fail if this factor were absent?
execution: subagent
prompt: ./prompt.md
input: artifact (string), factor (string), conclusion (string)
used-by: [counterfactual-probing]
---

# Necessity Evaluation

Scores probability of necessity: P(Y would not hold | X absent).

## Execution

Subagent — spawned via subagent-spawning/spawn-agent.

## Why Subagent

Necessity evaluation requires careful counterfactual reasoning about absence, isolated from sufficiency judgments.

## Input

- **artifact**: The original artifact
- **factor**: The factor to evaluate for necessity
- **conclusion**: The conclusion that may depend on this factor

## Output

- **pn_score**: Probability of necessity (0.0–1.0)
- **reasoning**: Why this factor is/isn't necessary
- **alternative_paths**: Other ways the conclusion could hold without this factor
- **confidence**: Confidence in the PN estimate

## Budget

One unit = one necessity evaluation per factor.

More from yogsoth-ai/de-anthropocentric-research-engine