instinct-system

Track recurring project patterns with confidence-scored instincts in .claude/instincts.md.

1|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/Maj3D10/Training-Platform --skill instinct-system-maj3d10
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: instinct-system
Source: https://github.com/Maj3D10/Training-Platform/tree/main/.agent/skills/instinct-system
Command: npx skills add https://github.com/Maj3D10/Training-Platform --skill instinct-system-maj3d10

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you consistently detect recurring project conventions without mistaking a single observation for a rule, by using confidence-scored hypotheses that mature only after confirmation.

Core Features & Use Cases

  • Confidence-scored instinct lifecycle: Starts at low confidence (hypothesis), increases through confirmations, and reaches promotion only after strong evidence.
  • Project-scoped learning and persistence: Stores instincts per project in .claude/instincts.md and promotes confirmed rules into MEMORY.md.
  • Observe-hypothesize-confirm discipline: Actively seeks confirming or disconfirming evidence in related files, preventing stagnation and overfitting to anomalies.
  • Session-start loading and staleness checks: Loads high-confidence instincts at the start of a session and flags stale entries for review.

Quick Start

Load the instinct-system skill at the start of a session to read and apply existing project instincts from .claude/instincts.md.

Frequently Asked Questions about instinct-system

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I track project conventions without mistaking a single observation for a rule?

To track project conventions safely, apply a confidence-scored instinct workflow that starts hypotheses at low confidence and promotes them only after confirming evidence across related files.

What is the best way to maintain codebase consistency across multiple development sessions?

Maintaining codebase consistency across sessions requires loading high-confidence instincts from project-scoped storage at session start, flagging stale entries, and applying confirmed architectural and naming patterns.

How do I validate naming and formatting patterns across a codebase?

Validate naming and formatting patterns by actively seeking confirming or disconfirming evidence in related files, applying an observe-hypothesize-confirm discipline to prevent overfitting to anomalies.

When should confirmed project instincts be promoted to permanent memory rules?

Confirmed project instincts should be promoted to permanent MEMORY.md rules only after they reach a 0.9 confidence score through multiple confirmations during software development sessions.

Does the instinct learning workflow require any specific dependencies or external libraries?

The instinct learning workflow requires no external dependencies, relying solely on maintaining project-scoped storage in local markdown files for persistence and session onboarding.

Why do false rules appear when detecting recurring architectural choices, and how can I prevent them?

False rules appear when single observations are mistaken for conventions; prevent this by using strict hypothesis testing with confidence updates that require strong evidence before promotion.