Version: v0.0

Influence finds repeated shaping patterns around a thought.

It owns one job:

detect what repeatedly moved attention, themes, or activity in a direction

Influence does not claim causality. It explains repeated patterns that may have shaped the context around a thought.

• Reads Capture snapshots today.
• Can later read Memory/Schema evidence at query time.
• Normalizes fragmented activity labels.
• Detects directional transitions such as A -> B.
• Computes count, P(B | A), P(B), lift, and confidence.
• Filters bidirectional noise.
• Reports source trails behind repeated patterns.
• Emits JSON, readable reports, graph lines, and DOT output.

A chain is kept only when it passes support and direction checks:

count(A -> B) >= 3
count(A) >= 5
P(B | A) > P(B)
A -> B is stronger than B -> A

Confidence:

count(A -> B) * (P(B | A) - P(B))

Prerequisites:

• Node.js 20+
• npm 10+

Install:

npm install

Validate:

npm run check

Run sample:

npm run sample

Analyze a Capture snapshot:

npm run analyze -- --input path\to\capture-snapshot.json --format report

All output formats:

npm run sample:all

• Influence is about repeated shaping, not first origin.
• It should avoid weak bidirectional patterns.
• It should cite source trails when available.
• It should never say a source caused a thought.

MIT. See LICENSE.