How do I install the Continuous Learning V2 skill?

Run npx skillmds add v1truv1us-ai-eng-system/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. Capability flags: docs only. 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, and the skill stays under its author's original license.

Who published Continuous Learning V2?

v1truv1us (@v1truv1us-ai-eng-system) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Continuous Learning V2 by v1truv1us-ai-eng-system

Name: Continuous Learning V2
Rating: 4.666666666666667 (3 reviews)
Author: v1truv1us

Continuous Learning V2

Instinct-based learning system with confidence scoring, import/export, and evolution into skills. The system learns patterns from sessions and promotes high-confidence instincts into durable skills. Use when extracting patterns from completed work.

v1truv1us f1d8be7 3.5 KB Updated Sep 10, 2026

File contents

Continuous Learning v2

Overview

A learning system that extracts patterns from completed sessions, scores them by confidence, and evolves high-confidence instincts into durable skills. Unlike v1's stop-hook pattern, v2 uses instinct-based learning with confidence scoring and evolution.

When to Use

After completing a significant feature or fix
When you've discovered a pattern worth remembering
When a session revealed a non-obvious solution
Before starting similar work to apply learned instincts

Instinct Lifecycle

1. Discovery

When a pattern is encountered during work:

Pattern: [What was discovered]
Context: [When this applies]
Confidence: [0.0-1.0 based on evidence]
Evidence: [What proves this works]

2. Validation

Before storing an instinct:

Pattern has been verified in at least one successful outcome
Context is specific enough to avoid false positives
Confidence score is justified with evidence
No conflicting instincts exist

3. Storage

Instincts are stored with:

instinct:
  name: "descriptive-name"
  pattern: "what to do"
  context: "when to apply"
  confidence: 0.85
  evidence: "why this works"
  created: "2026-01-15"
  usage_count: 3
  success_rate: 0.95

4. Evolution

When an instinct reaches thresholds:

Confidence > 0.9 AND usage_count > 5 AND success_rate > 0.9 → Promote to skill
Confidence < 0.5 AND usage_count > 3 → Deprecate
Confidence unchanged after 30 days → Review

Confidence Scoring

Score	Meaning	Action
0.9-1.0	Proven pattern, multiple successes	Promote to skill candidate
0.7-0.9	Strong pattern, some evidence	Use with confidence
0.5-0.7	Plausible pattern, limited evidence	Use cautiously, verify
0.3-0.5	Weak pattern, speculative	Note but don't rely on
0.0-0.3	Unproven, likely incorrect	Discard

Import/Export

Export Instincts

# Export all instincts to JSON
# Format: [{name, pattern, context, confidence, evidence}]

Import Instincts

# Import instincts from JSON
# Merge with existing, update confidence on duplicates

Anti-Rationalization Table

Excuse	Counter
"I'll remember this pattern"	Human memory is unreliable. Document it now with context and evidence.
"This is too specific to be useful"	Specific patterns become general skills through evolution. Start specific, generalize later.
"I don't have time to document"	2 minutes now saves 2 hours of rediscovery later. Use the instinct template.
"The confidence score is subjective"	Confidence is a starting point. Usage and success rates provide objective data over time.

Imported from continual-learning/continual-learning (MIT, cursor/plugins)

Continual Learning

Keep AGENTS.md current by delegating the memory update flow to one subagent.

Trigger

Use when the user asks to mine prior chats, maintain AGENTS.md, or run the continual-learning loop.

Workflow

Call agents-memory-updater.
Return the updater result.

Guardrails

Keep the parent skill orchestration-only.
Do not mine transcripts or edit files in the parent flow.
Do not bypass the subagent.

---name: continuous-learning-v2description: Instinct-based learning system with confidence scoring, import/export, and evolution into skills. The system learns patterns from sessions and promotes high-confidence instincts into durable skills. Use when extracting patterns from completed work.---# Continuous Learning v2## OverviewA learning system that extracts patterns from completed sessions, scores them by confidence, and evolves high-confidence instincts into durable skills. Unlike v1's stop-hook pattern, v2 uses instinct-based learning with confidence scoring and evolution.## When to Use- After completing a significant feature or fix- When you've discovered a pattern worth remembering- When a session revealed a non-obvious solution- Before starting similar work to apply learned instincts## Instinct Lifecycle### 1. DiscoveryWhen a pattern is encountered during work:```Pattern: [What was discovered]Context: [When this applies]Confidence: [0.0-1.0 based on evidence]Evidence: [What proves this works]```### 2. ValidationBefore storing an instinct:- [ ] Pattern has been verified in at least one successful outcome- [ ] Context is specific enough to avoid false positives- [ ] Confidence score is justified with evidence- [ ] No conflicting instincts exist### 3. StorageInstincts are stored with:```yamlinstinct: name: "descriptive-name" pattern: "what to do" context: "when to apply" confidence: 0.85 evidence: "why this works" created: "2026-01-15" usage_count: 3 success_rate: 0.95```### 4. EvolutionWhen an instinct reaches thresholds:- Confidence > 0.9 AND usage_count > 5 AND success_rate > 0.9 → Promote to skill- Confidence < 0.5 AND usage_count > 3 → Deprecate- Confidence unchanged after 30 days → Review## Confidence Scoring| Score | Meaning | Action ||-------|---------|--------|| 0.9-1.0 | Proven pattern, multiple successes | Promote to skill candidate || 0.7-0.9 | Strong pattern, some evidence | Use with confidence || 0.5-0.7 | Plausible pattern, limited evidence | Use cautiously, verify || 0.3-0.5 | Weak pattern, speculative | Note but don't rely on || 0.0-0.3 | Unproven, likely incorrect | Discard |## Import/Export### Export Instincts```bash# Export all instincts to JSON# Format: [{name, pattern, context, confidence, evidence}]```### Import Instincts```bash# Import instincts from JSON# Merge with existing, update confidence on duplicates```## Anti-Rationalization Table| Excuse | Counter ||--------|---------|| "I'll remember this pattern" | Human memory is unreliable. Document it now with context and evidence. || "This is too specific to be useful" | Specific patterns become general skills through evolution. Start specific, generalize later. || "I don't have time to document" | 2 minutes now saves 2 hours of rediscovery later. Use the instinct template. || "The confidence score is subjective" | Confidence is a starting point. Usage and success rates provide objective data over time. |## Imported from continual-learning/continual-learning (MIT, cursor/plugins)# Continual LearningKeep `AGENTS.md` current by delegating the memory update flow to one subagent.## TriggerUse when the user asks to mine prior chats, maintain `AGENTS.md`, or run the continual-learning loop.## Workflow1. Call `agents-memory-updater`.2. Return the updater result.## Guardrails- Keep the parent skill orchestration-only.- Do not mine transcripts or edit files in the parent flow.- Do not bypass the subagent.

Frequently asked questions