How do I install the Arc Learning skill?

Run npx skillmds@latest add tomevault-io/arc-learning 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 Arc Learning safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Arc Learning?

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 Arc Learning free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Arc Learning?

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

Arc Learning by tomevault-io — Agent Skill

Name: Arc Learning
Author: tomevault-io

Arc Learning

Use when you have accumulated instincts and want to cluster related ones into higher-level skills, commands, or agents. Use when instinct evolve suggests candidates. Use when you want to consolidate behavioral patterns into reusable abstractions.

tomevault-io Updated Sep 17, 2026

File contents

Instinct Clustering

Overview

Cluster related instincts into higher-level abstractions: skills, commands, or agents. This skill analyzes accumulated behavioral patterns (instincts) and identifies groups that can be consolidated into reusable components.

Position: observe → instincts → cluster (this skill) → skills/commands/agents

Quick Reference

Task	Command
Scan instincts	`node "${SKILL_ROOT}/scripts/learn.js" scan --project {p}`
Preview clusters	`node "${SKILL_ROOT}/scripts/learn.js" preview --project {p}`
Generate artifact	`node "${SKILL_ROOT}/scripts/learn.js" generate --cluster N --project {p} [--type skill\|command\|agent] [--name custom-name] [--dry-run]`
List evolved	`node "${SKILL_ROOT}/scripts/learn.js" list --project {p}`

Infrastructure Commands

Set SKILL_ROOT from skill loader header (# SKILL_ROOT: ...):

: "${SKILL_ROOT:=${ARCFORGE_ROOT:-}/skills/arc-learning}"
if [ ! -d "$SKILL_ROOT" ]; then
  echo "ERROR: SKILL_ROOT=$SKILL_ROOT does not exist. Set ARCFORGE_ROOT or SKILL_ROOT manually." >&2
  exit 1
fi

Workflow

Scan: Load all instincts from ~/.claude/instincts/{project}/ and global/
Cluster: Group by domain, then within each domain use trigger fingerprint similarity (Jaccard >= 0.6) to find sub-clusters
Filter: Only process clusters with 3+ instincts, at least 1 with confidence >= 0.6
Preview: Display candidate clusters with type recommendations and suggested names
Generate: Create skill, command, or agent from a cluster (auto-classified or user-overridden)
Track: Record evolution to ~/.claude/evolved/evolved.jsonl for deduplication

Note: Generated files are scaffolds — refine before deployment.

When to Use

instinct evolve suggests clustering candidates
5+ instincts accumulated in the same domain
User wants to consolidate behavioral patterns into reusable components
User explicitly asks to cluster or organize instincts

When NOT to Use

Fewer than 3 instincts in any domain
User wants to save a single pattern (use /recall instead)
User wants to reflect on diaries (use /reflect instead)
User wants to confirm/contradict individual instincts (use arc-observing)

Generation Types

Classification is hybrid: domain+threshold primary, keywords as tiebreaker.

Rule	Condition	Type
1	Domain is `workflow` or `automation` + avg confidence >= 0.7	command
2	Cluster size >= 3 + avg confidence >= 0.75	agent
3	Default (no primary rule matches)	skill

Tiebreaker: if behavioral tokens ("always", "prefer", "avoid") outnumber action tokens ("when starting", "when running") 2:1 → skill. Reverse → command.

Commands always produce a backing skill (arcforge convention). Source instincts are left in place after evolution.

Key Principles

User-driven: Preview clusters and let user decide what to create
Minimum cluster size: 3+ instincts required per cluster
Quality threshold: At least 1 instinct must have confidence >= 0.6
Source tracking: Record which instincts were evolved via JSONL log
No re-evolution: isAlreadyEvolved() prevents duplicate generation

Converted and distributed by TomeVault — claim your Tome and manage your conversions.

---name: arc-learningdescription: Use when you have accumulated instincts and want to cluster related ones into higher-level skills, commands, or agents. Use when instinct evolve suggests candidates. Use when you want to consolidate behavioral patterns into reusable abstractions.---# Instinct Clustering## OverviewCluster related instincts into higher-level abstractions: skills, commands, or agents. This skill analyzes accumulated behavioral patterns (instincts) and identifies groups that can be consolidated into reusable components.**Position:** `observe → instincts → cluster (this skill) → skills/commands/agents`## Quick Reference| Task | Command ||------|---------|| **Scan instincts** | `node "${SKILL_ROOT}/scripts/learn.js" scan --project {p}` || **Preview clusters** | `node "${SKILL_ROOT}/scripts/learn.js" preview --project {p}` || **Generate artifact** | `node "${SKILL_ROOT}/scripts/learn.js" generate --cluster N --project {p} [--type skill\|command\|agent] [--name custom-name] [--dry-run]` || **List evolved** | `node "${SKILL_ROOT}/scripts/learn.js" list --project {p}` |## Infrastructure Commands**Set SKILL_ROOT** from skill loader header (`# SKILL_ROOT: ...`):```bash: "${SKILL_ROOT:=${ARCFORGE_ROOT:-}/skills/arc-learning}"if [ ! -d "$SKILL_ROOT" ]; then echo "ERROR: SKILL_ROOT=$SKILL_ROOT does not exist. Set ARCFORGE_ROOT or SKILL_ROOT manually." >&2 exit 1fi```## Workflow1. **Scan**: Load all instincts from `~/.claude/instincts/{project}/` and `global/`2. **Cluster**: Group by domain, then within each domain use trigger fingerprint similarity (Jaccard >= 0.6) to find sub-clusters3. **Filter**: Only process clusters with 3+ instincts, at least 1 with confidence >= 0.64. **Preview**: Display candidate clusters with type recommendations and suggested names5. **Generate**: Create skill, command, or agent from a cluster (auto-classified or user-overridden)6. **Track**: Record evolution to `~/.claude/evolved/evolved.jsonl` for deduplication**Note:** Generated files are scaffolds — refine before deployment.## When to Use- `instinct evolve` suggests clustering candidates- 5+ instincts accumulated in the same domain- User wants to consolidate behavioral patterns into reusable components- User explicitly asks to cluster or organize instincts## When NOT to Use- Fewer than 3 instincts in any domain- User wants to save a single pattern (use /recall instead)- User wants to reflect on diaries (use /reflect instead)- User wants to confirm/contradict individual instincts (use arc-observing)## Generation TypesClassification is hybrid: domain+threshold primary, keywords as tiebreaker.| Rule | Condition | Type ||------|-----------|------|| 1 | Domain is `workflow` or `automation` + avg confidence >= 0.7 | **command** || 2 | Cluster size >= 3 + avg confidence >= 0.75 | **agent** || 3 | Default (no primary rule matches) | **skill** |Tiebreaker: if behavioral tokens ("always", "prefer", "avoid") outnumber action tokens ("when starting", "when running") 2:1 → skill. Reverse → command.Commands always produce a backing skill (arcforge convention). Source instincts are left in place after evolution.## Key Principles- **User-driven**: Preview clusters and let user decide what to create- **Minimum cluster size**: 3+ instincts required per cluster- **Quality threshold**: At least 1 instinct must have confidence >= 0.6- **Source tracking**: Record which instincts were evolved via JSONL log- **No re-evolution**: `isAlreadyEvolved()` prevents duplicate generation---> Converted and distributed by [TomeVault](https://tomevault.io/claim/gregoryho) — claim your Tome and manage your conversions.

Frequently asked questions