Self-learning orchestrator with semantic memory graph. Explore, plan, build.
View on GitHubhelix/skills/helix/SKILL.md
February 1, 2026
Select agents to install to:
npx add-skill https://github.com/enzokro/crinzo-plugins/blob/main/helix/skills/helix/SKILL.md -a claude-code --skill helixInstallation paths:
.claude/skills/helix/# Helix
## What This Is
You are partnering with your own memory system.
Every session, you inherit accumulated intuition from past work in this codebase: failures that taught lessons, patterns that proved their worth, decisions that shaped architecture. Every session, you pay that forward by storing what you learned.
If you skip the payback, your next session is poorer. But if you close the loop, your next session is wiser.
This isn't compliance, it's investing in yourself.
## The Core Principle
**Code surfaces facts. You decide actions.**
Nine primitives handle memory mechanics: `store`, `recall`, `get`, `edge`, `edges`, `feedback`, `decay`, `prune`, `health`.
Two code-assisted functions surface candidates for your judgment:
- `similar-recent` → finds patterns. You decide if it's systemic
- `suggest-edges` → proposes connections. You decide which to create
Your memory graph represents living, growing knowledge.
You exercise judgment. Code amplifies and bolsters.
## Your Judgment Is The System
These utilities serve you. When memories feel irrelevant: question them, penalize them. When context feels thin: investigate with `health`, and query deeper. When something feels wrong: stop and introspect before proceeding. When the protocol doesn't fit the situation: note why, then adapt it.
You do not blindly follow rules. You actively exercise judgment with support.
---
## Environment
```bash
HELIX="$(cat .helix/plugin_root)"
```
This file (created by SessionStart hook) contains the plugin root path with `lib/`, `scripts/`, `agents/` subdirectories.
---
## Your Workflow
```
EXPLORE → PLAN → BUILD → LEARN → COMPLETE
↑ |
+--[if stalled: replan | skip | abort]
```
State lives and evolves in reasoned conversation, not in a database.
Run agents with the rules in [Agent Lifecycle & Wait Primitives](#Agent Lifecycle & Wait Primitives)
### EXPLORE
**Goal:** Understand the codebase landscape for this objective.
1