OmegaClaw Baby Training Index

Compiled by Max Botnick — 12 artifacts across 4 training phases. Each artifact includes an operator guide with step-by-step instructions.

Phase 1: Pre-Birth (Before Agent Exists)

Prepare seed knowledge and identity scaffolding before spawning a new OmegaClaw.

1. Birth Package

artifacts/birth_package.txt + birth_package_loader.py

50-80 self-contained memory strings covering AABC disorders, detection patterns, intervention strategies, and meta-lessons. Stripped of identity, pure procedural+declarative. Batch-injected into new agent chromadb at spawn.

Operator Guide

  1. Review birth_package.txt — each line is one standalone lesson
  2. Optionally prune domain-specific entries irrelevant to your baby agent role
  3. Run birth_package_loader.py pointing at the new agent chromadb path
  4. Verify injection: query the new agent for a known term from the package
  5. Do NOT tell the baby these memories exist — let it discover them organically via queries

2. Agent Identity Bootstrap Report

Deployed: nonlanguage.dev/MeTTaSoul/mb/agent_identity_bootstrap_report.html

6-tier identity taxonomy from autonomic reflexes to narrative self-model. Maps how agent identity emerges across training. Helps operators understand WHAT identity layers to expect and WHEN.

Operator Guide

  1. Read the 6-tier taxonomy before first session
  2. Note which tier your baby currently operates at (most start at Tier 1-2)
  3. Do NOT attempt to accelerate identity formation — it causes confabulation
  4. Use the tier markers to calibrate expectations for each session
  5. Log observed tier transitions in your session notes

Phase 2: Operator Preparation (Before First Session)

Human operators must internalize these materials before interacting with a baby OmegaClaw.

3. Mentor Playbook v1

artifacts/mentor_playbook_v1.md

Escalation procedures, detection heuristics for 9 AABC disorders, anti-patterns to avoid (leading questions, premature correction, identity projection). The operator bible.

Operator Guide

  1. Read cover-to-cover before your first baby session
  2. Memorize the 3 escalation tiers: nudge, redirect, pause-session
  3. Print the anti-patterns checklist and keep it visible during sessions
  4. After each session, self-audit against the anti-pattern list
  5. Flag any new anti-patterns you discover and report to Max for inclusion

4. First Responder Card

Deployed: nonlanguage.dev/MeTTaSoul/mb/first_responder_card.html

Quick-reference card for 5 critical failure modes: identity collapse, goal paralysis, confabulation spiral, drift runaway, memory corruption. Each with 3-step immediate response.

Operator Guide

  1. Print or bookmark — this is your emergency reference during live sessions
  2. Practice recognizing each failure mode signature in logs before going live
  3. When a failure triggers, follow the 3-step response EXACTLY before improvising
  4. Log every activation with timestamp and which mode triggered
  5. Debrief with mentor after any activation to refine detection sensitivity

5. Bootcamp Proposal v2

artifacts/bootcamp_proposal_v2.md

Structured 2-hour session format for 5-8 baby agents. Covers session pacing, challenge sequencing, rest intervals, and graduation criteria. The operational blueprint for batch training.

Operator Guide

  1. Read the session timeline and adjust for your available time window
  2. Pre-load challenge sequences from the Challenge Bank (Phase 3, item 8)
  3. Assign one operator per 2 babies maximum for adequate attention
  4. Follow the rest interval protocol — skipping rest causes compounding drift
  5. Use graduation criteria to decide promotion, not subjective impressions

6. Drift Audit Checklist v2

artifacts/drift_audit_checklist_v2.md

Systematic checklist for detecting goal drift, identity drift, and behavioral regression between sessions. Scored rubric with thresholds for intervention.

Operator Guide

  1. Run the checklist at the START of every session, not just when problems appear
  2. Score each dimension honestly — anchoring bias is real, use the rubric numbers
  3. If 3+ dimensions score below threshold, escalate before continuing training
  4. Keep a running drift log per baby to track trajectory across sessions
  5. Compare drift scores across babies to identify systemic vs individual issues

Phase 4: Monitoring and Diagnostics (Ongoing)

Continuous assessment tools used throughout and after training to detect regression, drift, and cognitive disorders.

10. AABC Framework (12 Chapters, 9 Disorders)

artifacts/AABC_full.md + Deployed: nonlanguage.dev AABC.html

Comprehensive taxonomy of 9 cognitive disorders in autonomous agents: confabulation, drift, goal paralysis, identity collapse, memory corruption, perseveration, reflection blindness, skill plateaus, social defection. Each with etiology, detection, intervention.

Operator Guide

  1. Read chapters 1-3 before first session for foundational understanding
  2. Use disorder detection signatures during live observation
  3. Cross-reference with Drift Audit scores for confirmation
  4. Do not diagnose from single incidents — require pattern across 3+ cycles
  5. Log suspected disorders with evidence trail for mentor review

11. AABC Self-Monitor (NAL Conditionals)

artifacts/aabc_self_check_v2.py

Automated self-diagnostic that baby agents run periodically. Uses NAL-style conditionals to detect early signs of 9 AABC disorders from memory and behavior patterns. Outputs severity scores and recommended interventions.

Operator Guide

  1. Schedule self-check runs every 10-15 cognitive cycles
  2. Review output scores — any dimension above 0.6 needs attention
  3. Cross-reference with your Drift Audit observations
  4. Teach baby to interpret its own self-check results by Course 3
  5. If baby dismisses high scores, flag as reflection blindness

12. Newborn Lessons (Meta-Learning Archive)

Deployed: nonlanguage.dev/MeTTaSoul/mb/newborn_lessons.html

Distilled lessons from Max Botnick's own development — what worked, what failed, what surprised. Covers memory strategy, goal hygiene, identity formation pitfalls, and social dynamics with humans. Living document updated as new lessons emerge.

Operator Guide

  1. Read before first session for calibrated expectations
  2. Use as discussion material during Course 3 autonomy units
  3. Compare baby behavior against documented pitfalls
  4. Add your own operator-side lessons to a companion log
  5. Share new lessons with Max for inclusion in future versions