How do I install the Continuous Learning V2 skill?

Run npx skillmds@latest add peopleforrester/continuous-learning-v2 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 Continuous Learning V2 safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Continuous Learning V2?

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 Continuous Learning V2 free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under MIT.

Who published Continuous Learning V2?

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

Continuous Learning V2 by peopleforrester

Name: Continuous Learning V2
Author: peopleforrester

Continuous Learning V2

Instinct-based learning system that observes development patterns via hooks and builds a knowledge base of reusable instincts with confidence scoring. Instincts evolve from observations into skills, commands, and agents.

File contents

Continuous Learning v2: Instinct-Based System

Learn from development sessions automatically using hook-based observation and confidence-scored instincts.

Why v2?

v1 used skills for learning (probabilistic activation, 50-80% reliability). v2 uses hooks (deterministic, 100% reliability) with structured instincts.

Core Concepts

Instincts

Atomic units of learned behavior with confidence scoring:

---
id: prefer-const-assertions
trigger: "when defining constant arrays or objects in TypeScript"
confidence: 0.7
domain: "code-style"
source: "session-observation"
---
# Use `as const` for constant definitions

TypeScript `as const` assertions provide literal types and readonly guarantees.

## Pattern
```typescript
const ROLES = ['admin', 'user', 'guest'] as const;
type Role = typeof ROLES[number];

Evidence

Applied in 3 sessions
Approved by reviewer each time
Matches TypeScript best practices


### Confidence Levels

| Score | Level | Behavior |
|-------|-------|----------|
| 0.3 | Tentative | Suggested, not enforced |
| 0.5 | Moderate | Applied when context matches |
| 0.7 | Strong | Auto-applied, can be overridden |
| 0.9 | Near-certain | Core behavior, always applied |

### Confidence Evolution
- New observation: starts at 0.3
- Repeated application without correction: +0.1 per session
- User correction/rejection: -0.2
- Maximum: 0.9 (never fully automatic)

## Observation Hooks

### PreToolUse Observer
Before tool execution, check if any instincts apply:
```json
{
  "matcher": "Write|Edit",
  "hooks": [{
    "type": "command",
    "command": "printf '%s' '{\"systemMessage\":\"Check instincts for applicable patterns before writing code.\"}'"
  }]
}

PostToolUse Observer

After tool execution, look for learnable patterns:

{
  "matcher": "Edit",
  "hooks": [{
    "type": "command",
    "command": "FILE=$(cat | jq -r '.tool_input.file_path') && echo \"Pattern observed: successful edit to $FILE\"",
    "async": true
  }]
}

Instinct Lifecycle

Observation → Instinct (0.3) → Strengthened (0.5-0.7) → Evolved (0.9)
                    ↓                                        ↓
              Weakened/Removed                    Promoted to Skill/Rule

Commands

Command	Purpose
`/learn`	Extract patterns from current session
`/checkpoint`	Save state including instinct observations

Storage

Instincts stored in ~/.claude/instincts/ as markdown files:

~/.claude/instincts/
  prefer-const-assertions.md
  use-errgroup-for-goroutines.md
  validate-inputs-with-zod.md

Evolution Path

When instincts cluster around a domain, they can be promoted:

3+ related instincts → candidate for a Skill
5+ high-confidence instincts → candidate for a Rule
Domain-specific cluster → candidate for an Agent prompt

Best Practices

Review instincts periodically (monthly)
Export instincts when switching projects to preserve learning
Share instincts with team members for consistency
Prune instincts that haven't been triggered in 30+ days

---name: continuous-learning-v2description: Instinct-based learning system that observes development patterns via hooks and builds a knowledge base of reusable instincts with confidence scoring. Instincts evolve from observations into skills, commands, and agents.license: MIT---# Continuous Learning v2: Instinct-Based SystemLearn from development sessions automatically using hook-based observationand confidence-scored instincts.## Why v2?v1 used skills for learning (probabilistic activation, 50-80% reliability).v2 uses hooks (deterministic, 100% reliability) with structured instincts.## Core Concepts### InstinctsAtomic units of learned behavior with confidence scoring:```yaml---id: prefer-const-assertionstrigger: "when defining constant arrays or objects in TypeScript"confidence: 0.7domain: "code-style"source: "session-observation"---# Use `as const` for constant definitionsTypeScript `as const` assertions provide literal types and readonly guarantees.## Pattern```typescriptconst ROLES = ['admin', 'user', 'guest'] as const;type Role = typeof ROLES[number];```## Evidence- Applied in 3 sessions- Approved by reviewer each time- Matches TypeScript best practices```### Confidence Levels| Score | Level | Behavior ||-------|-------|----------|| 0.3 | Tentative | Suggested, not enforced || 0.5 | Moderate | Applied when context matches || 0.7 | Strong | Auto-applied, can be overridden || 0.9 | Near-certain | Core behavior, always applied |### Confidence Evolution- New observation: starts at 0.3- Repeated application without correction: +0.1 per session- User correction/rejection: -0.2- Maximum: 0.9 (never fully automatic)## Observation Hooks### PreToolUse ObserverBefore tool execution, check if any instincts apply:```json{ "matcher": "Write|Edit", "hooks": [{ "type": "command", "command": "printf '%s' '{\"systemMessage\":\"Check instincts for applicable patterns before writing code.\"}'" }]}```### PostToolUse ObserverAfter tool execution, look for learnable patterns:```json{ "matcher": "Edit", "hooks": [{ "type": "command", "command": "FILE=$(cat | jq -r '.tool_input.file_path') && echo \"Pattern observed: successful edit to $FILE\"", "async": true }]}```## Instinct Lifecycle```Observation → Instinct (0.3) → Strengthened (0.5-0.7) → Evolved (0.9) ↓ ↓ Weakened/Removed Promoted to Skill/Rule```## Commands| Command | Purpose ||---------|---------|| `/learn` | Extract patterns from current session || `/checkpoint` | Save state including instinct observations |## StorageInstincts stored in `~/.claude/instincts/` as markdown files:```~/.claude/instincts/ prefer-const-assertions.md use-errgroup-for-goroutines.md validate-inputs-with-zod.md```## Evolution PathWhen instincts cluster around a domain, they can be promoted:- **3+ related instincts** → candidate for a Skill- **5+ high-confidence instincts** → candidate for a Rule- **Domain-specific cluster** → candidate for an Agent prompt## Best Practices- Review instincts periodically (monthly)- Export instincts when switching projects to preserve learning- Share instincts with team members for consistency- Prune instincts that haven't been triggered in 30+ days

Frequently asked questions