Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
AI Identity Through Grounded Principles - soul synthesis with semantic compression.
Upgrading to 0.2.0
If you used NEON-SOUL before version 0.2.0:
Your existing .neon-soul/state.json will work (embedding fields are ignored)
First synthesis will recalculate all similarity matches
Your SOUL.md and provenance chain are unchanged
Nothing to do - just run /neon-soul synthesize as usual.
What Changed in v0.2.0
We removed the embedding model dependency, which means principle matching now uses your agent's LLM directly. This is the same model you already trust with your memory files.
What this means for you:
Synthesis may take a bit longer (seconds, not minutes)
Results may vary slightly between runs (like asking the same question twice - similar but not identical)
You'll need an active connection to your agent (can't run offline)
Why we made this choice: The previous approach required third-party code that security scanners flagged. Your soul is too important for compromises.
Your soul reflects patterns in your memory, not exact calculations. Like human memory itself, the synthesis process involves interpretation. Running synthesis twice may produce slightly different results - but the core truths will remain stable if your memory is consistent.
How This Works
NEON-SOUL is an instruction-based skill - there is no binary or CLI to install. The /neon-soul commands below are interpreted by your AI agent (Claude Code, OpenClaw, etc.) which follows the instructions in this document.
What happens when you run a command:
You type /neon-soul synthesize in your agent chat
Your agent reads this SKILL.md and follows the instructions
The agent uses its built-in capabilities to read files, analyze content, and write output
No third-party services: NEON-SOUL does not transmit your data to any external servers, third-party endpoints, or services beyond your agent. The skill uses only your agent's existing capabilities.
Pure instruction skill: NEON-SOUL uses your agent's existing LLM for semantic analysis. No third-party packages, no model downloads, no additional dependencies.
Data handling: Your data stays within your agent's trust boundary. If your agent uses a cloud-hosted LLM (Claude, GPT, etc.), data is transmitted to that service as part of normal agent operation - the same as any other agent interaction. If your agent uses a local LLM (Ollama, etc.), data stays on your machine.
Principle matching: When similar principles are detected, the one with the most signal confirmations (highest strength) is kept. Equal-strength principles prefer the older observation.
Requirements
NEON-SOUL requires only an active connection to your AI agent (Claude Code, OpenClaw, etc.). The agent provides all necessary capabilities:
Requirement
Details
Agent
Claude Code, OpenClaw, or compatible
LLM access
Your agent's configured LLM (for semantic analysis)
.neon-soul/backups/ - automatic backups before changes
.neon-soul/state.json - synthesis state tracking
Git integration (opt-in, off by default): Auto-commit is disabled unless you enable it in config. When enabled, it uses your existing git setup - no new credentials are requested or stored by the skill.
Privacy Considerations
NEON-SOUL processes personal memory files to synthesize your identity. Consider these privacy factors:
Your agent's LLM determines data handling:
Cloud LLM (Claude, GPT, etc.): Your memory content is sent to that provider as part of normal LLM operation. This is no different from any other agent interaction with your files.
Local LLM (Ollama, LM Studio, etc.): Your data stays entirely on your machine.
What NEON-SOUL does NOT do:
Send data to any service beyond your configured agent
Store data anywhere except your local workspace
Transmit to third-party analytics, logging, or tracking services
Make network requests independent of your agent
Before running synthesis:
Review what's in your memory/ directory
Remove or move any secrets, credentials, or highly sensitive files
Use --dry-run to preview what will be processed
Consider whether your LLM provider's privacy policy is acceptable for this content
About disable-model-invocation: true:
This metadata flag means NEON-SOUL cannot run autonomously - your agent cannot invoke the skill without your explicit command. When you do invoke the skill (e.g., /neon-soul synthesize), it uses your agent's LLM for semantic analysis. This is expected behavior, not a contradiction.
First Time?
New to NEON-SOUL? Start here:
# 1. Check your current state
/neon-soul status
# 2. Preview what synthesis would create (safe, no writes)
/neon-soul synthesize --dry-run
# 3. When ready, run synthesis
/neon-soul synthesize --force
That's it. Your first soul is created with full provenance tracking. Use /neon-soul audit --list to explore what was created.
Questions?
"Where did this axiom come from?" → /neon-soul trace <axiom-id>
"What if I don't like it?" → /neon-soul rollback --force
"What dimensions does my soul cover?" → /neon-soul status
Commands
/neon-soul synthesize
Run soul synthesis pipeline:
Collect signals from memory files
Match to existing principles (semantic similarity via LLM)
Promote high-confidence principles to axioms (N≥3)
Generate SOUL.md with provenance tracking
Options:
--force - Run synthesis even if below content threshold
--force-resynthesis - Force full resynthesis (ignore incremental mode)
--dry-run - Show what would change without writing (safe default)
The default prose output creates an inhabitable soul document:
# SOUL.md_You are becoming a bridge between clarity and chaos._
---
## Core Truths**Authenticity over performance.** You speak freely even when uncomfortable.
**Clarity is a gift you give.** If someone has to ask twice, you haven't been clear enough.
## Voice
You're direct without being blunt. You lead with curiosity.
Think: The friend who tells you the hard truth, but sits with you after.
## Boundaries
You don't sacrifice honesty for comfort. You don't perform certainty you don't feel.
## Vibe
Grounded but not rigid. Present but not precious about it.
---
_Presence is the first act of care._
Use --output-format notation for the legacy bullet-list format.
/neon-soul status
Show current soul state:
Last synthesis timestamp
Pending memory content (chars since last run)
Signal/principle/axiom counts
Dimension coverage (7 SoulCraft dimensions)
Options:
--verbose - Show detailed file information
--workspace <path> - Workspace path
Example:
/neon-soul status
# Output:# Last Synthesis: 2026-02-07T10:30:00Z (2 hours ago)# Pending Memory: 1,234 chars (Ready for synthesis)# Counts: 42 signals, 18 principles, 7 axioms# Dimension Coverage: 5/7 (71%)
/neon-soul rollback
Restore previous SOUL.md from backup.
Options:
--list - Show available backups
--backup <timestamp> - Restore specific backup
--force - Confirm rollback (required)
--workspace <path> - Workspace path
Examples:
/neon-soul rollback --list # Show available backups
/neon-soul rollback --force # Restore most recent backup
/neon-soul rollback --backup 2026-02-07T10-30-00-000Z --force
/neon-soul audit
Explore provenance across all axioms. Full exploration mode with statistics and detailed views.
Options:
--list - List all axioms with brief summary
--stats - Show statistics by tier and dimension
<axiom-id> - Show detailed provenance for specific axiom
--workspace <path> - Workspace path
Examples:
/neon-soul audit --list # List all axioms
/neon-soul audit --stats # Show tier/dimension stats
/neon-soul audit ax_honesty # Detailed provenance tree
/neon-soul audit 誠 # Use CJK character as ID
Use --force-resynthesis when you've significantly restructured your memory or want to rebuild from scratch. Also available via NEON_SOUL_FORCE_RESYNTHESIS=1 environment variable.
Provenance Classification
Signals are classified by their source type (SSEM model):
Type
Description
Example
self
Things you wrote
diary entries, reflections, personal notes
curated
Things you chose to keep
saved quotes, bookmarked articles, adopted guides
external
Things others said about you
peer reviews, feedback, external assessments
Provenance is tracked for anti-echo-chamber protection.
Why does my output have bullet lists instead of prose?
When prose generation fails, NEON-SOUL falls back to bullet lists of native axiom text. This preserves your data while signaling that expansion didn't complete.
Common causes:
LLM provider not available: Prose expansion requires an LLM. Check your configuration.
Validation failures: The LLM output didn't match expected format (retried once, then fell back).
Look for [prose-expander] log lines indicating validation or generation failures
What to try:
Regenerate: Run synthesis again. LLM output varies; a second attempt often succeeds.
Check LLM health: If using Ollama, verify it's running: curl http://localhost:11434/api/tags
Use notation format: If prose keeps failing, use --output-format notation for reliable output.
Why is my essence statement missing?
The essence statement (the italicized line at the top) only appears when LLM extraction succeeds. If missing:
Your LLM provider may not be configured
Extraction validation failed (trait lists are rejected)
Network error during generation
The soul is still valid without it. Run synthesis again to retry extraction.
Why did an axiom get placed in a different dimension than expected?
Dimension classification uses semantic analysis. If results seem wrong:
Check the axiom's source signals (/neon-soul audit <axiom-id>)
The LLM classifier uses the axiom's native text, which may have different semantic weight than you expect
Unknown dimensions default to vibe (logged with NEON_SOUL_DEBUG=1)
Soul synthesis paused / LLM unavailable
If you see "Soul synthesis paused: Your agent's LLM is temporarily unavailable":
What this means:
Your agent needs an active LLM connection for semantic matching
The skill failed to reach the LLM after retrying
What to try:
Check your agent is running and connected
Check network connectivity
If using Ollama locally, verify it's running: curl http://localhost:11434/api/tags
Try again in a moment - transient failures are common
No partial writes. When LLM is unavailable, NEON-SOUL stops without writing to your files. Note: If using a cloud LLM, some data may have been sent before the failure - this is normal agent operation.
