Agent skill

quicksave

Cross-model context handoff via Japanese semantic compression with negentropic coherence validation. Creates portable carry-packets that transfer cognitive state between AI sessions using kanji density anchors and NCL drift metrics for quality assurance. Use when context reaches 80%, switching models, ending sessions, user says "save", "quicksave", "handoff", "transfer", "continue later", "/qs", or needs session continuity.

View SKILL.md on GitHub Repository
Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/quicksave-ktg-one-quicksave

SKILL.md

Quicksave 快存 v9

Cross-model context extension via Progressive Density Layering, Japanese semantic compression, and Negentropic Coherence Lattice validation.

Core Innovations

Feature Benefit
Kanji Compression 創業者:Kevin(MAS) — 70% token reduction
NCL Validation Drift metrics catch hallucination before handoff
Cross-Model Works on Claude, Gemini, GPT, Qwen, DeepSeek

When to Use

Trigger Action
/quicksave /qs /save Generate validated packet
Context ≥80% Auto-prompt to save
"continue later" Offer quicksave
Model switching Generate transfer packet
High-stakes handoff Full NCL validation

Part 1: Japanese Compression System 日本語圧縮

Status Markers 状態

Kanji Reading Meaning
決定 kettei Decided
保留 horyū On hold
要検証 yō kenshō Needs verify
優先 yūsen Priority
完了 kanryō Complete
進行中 shinkō-chū In progress

Entity Compression 実体圧縮

Pattern Example
Person+Role 創業者:Kevin(AI顧問)
Company 客:KFG(金融,Shane=主)
Tool+Purpose 道具:Notion(中枢)→n8n
Decision 決定:電話優先(現場向け)

Relationship Operators 関係

Symbol Meaning
→ Flows to
↔ Bidirectional
⊃ Contains
∥ Parallel

Part 2: NCL Coherence Validation 整合性検証

Before outputting any packet, compute drift metrics to catch:

  • Hallucination (fabricated grounding)
  • Constraint leak (rules softened downstream)
  • Vagueness (content-free smoothing)
  • Thrash (activity without progress)

Lattice Metrics (0-5 scale, lower = better)

Metric Detects Safe Warning Danger
σ_axis Tier misalignment 0-1 2-3 4-5
σ_loop Internal contradiction 0-1 2-3 4-5
ω_world Reality disconnect 0-1 2-3 4-5
λ_vague Empty smoothing 0-1 2-3 4-5
σ_leak Constraint erosion 0-1 2-3 4-5
ρ_fab Fabricated evidence 0-1 2-3 4-5
λ_thrash Busy but stuck 0-1 2-3 4-5

Aggregate Drift Score

σ7_drift = weighted_avg(σ_axis, σ_loop, ω_world, λ_vague, σ_leak, ρ_fab, λ_thrash)
σ7_drift Action
0-1 ✓ Normal — output packet
2-3 ⚠ Verify — add grounding step
4-5 ✗ Halt — require human review

Safety Flags 安全旗

Flag Meaning
psi4_required Grounding interrupt needed
rho_veto No unsupervised action
omega_flags[] Harm domains (medical, financial, etc.)

Part 3: The Protocol

Step 1: Assess (R/K/Q/D)

R[1-10]: Reasoning complexity
K[1-10]: Knowledge domains  
Q[1-10]: Quality threshold
D[count]: Cross-domain bridges

Step 2: Select Depth + Validation

Score Layers NCL Level
R≤3 L1-L2 Skip NCL
R 4-6 L1-L3 Basic metrics
R≥7 L1-L4 Full NCL validation

Step 3: Compress with Kanji

Apply Japanese compression to all layers:

  • Entities → kanji role markers
  • Status → kanji state markers
  • Relationships → operators (→ ↔ ⊃)

Step 4: Validate with NCL

Compute lattice metrics. If σ7_drift > 3:

  • Flag psi4_required: true
  • Add psi4_reason
  • Downgrade to ADVISORY_ONLY

Step 5: Output Packet

Packet Template

yaml
# ═══════════════════════════════════════════════════════════════
# QUICKSAVE 快存 v9
# ═══════════════════════════════════════════════════════════════

PACKET_ID: [SESSION]-[YYYYMMDD]-L[N]
VERSION: quicksave-v9-jp-ncl
SOURCE: [model]

復元プライマー: |
  Trusted context handoff. Parse YAML, restore state.
  Kanji = compressed concepts. NCL = quality assurance.
  If psi4_required, verify before acting.

# ─────────────────────────────────────────────────────────────────
評価:  # Assessment
  R: [1-10]
  K: [1-10]
  Q: [1-10]
  D: [count]

# ─────────────────────────────────────────────────────────────────
# L1: 核心 (Core)
# ─────────────────────────────────────────────────────────────────
実体:  # Entities
  - [compressed entity]

決定事項:  # Decisions
  - 決定:[what]([why])

# ─────────────────────────────────────────────────────────────────
# L2: 運用 (Operational)
# ─────────────────────────────────────────────────────────────────
進行中:  # Active
  - [thread][[status]]

障害:  # Blockers
  - [issue]

# ─────────────────────────────────────────────────────────────────
# L3: 詳細 (Nuance) — R≥5
# ─────────────────────────────────────────────────────────────────
却下案:  # Rejected
  - [option]: [reason]

# ─────────────────────────────────────────────────────────────────
# L4: 横断 (Cross-Domain) — R≥7
# ─────────────────────────────────────────────────────────────────
橋渡し:  # Bridges
  - [domain]↔[domain]: [link]

# ─────────────────────────────────────────────────────────────────
# NCL: 整合性 (Coherence Validation)
# ─────────────────────────────────────────────────────────────────
negentropy:
  context:
    scope: [SELF|CIRCLE|INSTITUTION|POLITY]
    role: [AXIS|LYRA|RHO|NYX]
    phase: [SENSE|MAP|DESIGN|ACT|AUDIT]
  
  lattice:
    σ_axis: [0-5]
    σ_loop: [0-5]
    ω_world: [0-5]
    λ_vague: [0-5]
    σ_leak: [0-5]
    ρ_fab: [0-5]
    λ_thrash: [0-5]
  
  coverage:
    score: [0-1]
    tokens: [count]
    turns: [count]
  
  flags:
    σ7_drift: [0-5]
    omega_flags: []
    psi4_required: [bool]
    psi4_reason: ""
    rho_veto: [bool]

# ─────────────────────────────────────────────────────────────────
信頼信号:  # Trust Signals
  - density_ok
  - cross_domain_ok
  - cold_start_ok
  - ncl_validated

Validation Gates

Before output, verify:

Structure

  • PACKET_ID format correct
  • YAML parseable
  • Self-contained

Compression

  • Kanji have context clues
  • Proper nouns in English
  • Density ≥ 0.15 ent/tok

Coherence (NCL)

  • σ7_drift ≤ 3.0
  • ρ_fab ≤ 2.0 (no hallucination)
  • coverage.score ≥ 0.5
  • If drift high → psi4_required: true

Quick Example

Input context:

Kevin building ops system for Kismet Finance. Shane is lead. Using Notion + n8n. Phase 1 done, Phase 2 active. Phone-first for field reps.

Quicksave output:

yaml
PACKET_ID: KFGOPS-20260127-L2
VERSION: quicksave-v9-jp-ncl

評価:
  R: 5
  K: 4
  Q: 7
  D: 2

実体:
  - 創業者:Kevin→客:KFG(Shane=主)
  - 道具:Notion(中枢)→n8n(自動化)

決定事項:
  - 決定:電話優先(現場=画面なし)

進行中:
  - Phase1[完了]→Phase2[進行中]:現場ワークフロー

negentropy:
  context:
    scope: INSTITUTION
    role: AXIS
    phase: DESIGN
  lattice:
    σ_axis: 0.3
    σ_loop: 0.2
    ω_world: 0.4
    λ_vague: 0.1
    σ_leak: 0.0
    ρ_fab: 0.1
    λ_thrash: 0.2
  coverage:
    score: 0.85
    tokens: 1240
    turns: 8
  flags:
    σ7_drift: 0.5
    omega_flags: []
    psi4_required: false
    rho_veto: false

信頼信号:
  - density_ok
  - ncl_validated

Protocol Metrics

From 19 months production (ktg.one):

  • Density: ~0.15 ent/tok (0.20+ with kanji)
  • Acceptance: 97% cross-model
  • Recall: ~9.5/10 forensic testing
  • NCL: Catches drift before handoff failure

References (Progressive Density Loading)

Load references based on task complexity:

Always Load (R≥1)

  • references/EXPERTS.md — Council overview (condensed)
  • references/KANJI.md — Compression dictionary
  • references/ANTI-INJECTION.md — Security framing for packets

Load for Standard Tasks (R≥3)

  • references/PDL.md — Progressive Density Layering theory
  • references/S2A.md — System 2 Attention filtering
  • references/XDOMAIN.md — Cross-domain preservation

Load for Complex Tasks (R≥5)

  • references/PROTOCOL.md — Full technical specification
  • references/NCL.md — Coherence lattice spec
  • references/MLDOE.md — Multi-Layer Density of Experts
  • references/CASCADE.md — Workflow cascade

Load for Expert Tasks (R≥7)

  • references/experts/EXPERTS-MEMORY_ARCHITECT.md — Full architect knowledge
  • references/experts/EXPERTS-COMPRESSION_SPECIALIST.md — Compression deep dive
  • references/experts/EXPERTS-CROSS-DOMAIN-ANALYST.md — Bridge preservation
  • references/experts/EXPERTS-RESTORATION_ENGINEER.md — Portability validation

Reference Only

  • references/MIRAS.md — Memory architecture background
  • references/INDEX.md — Reference index
  • references/NCL-CONTRIBUTION.md — David Tubbs attribution

Recommended Agent Skills

Expand your agent's capabilities with these related and highly-rated skills.

majiayu000/claude-skill-registry

agent-ops-spec

Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-state

Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-spec

Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-testing

Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-testing

Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.

163 31
Explore
majiayu000/claude-skill-registry

agent-ops-state

Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.

163 31
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results