Skip to content

memory-checkpoint.md

Latest commit

 

History

History
220 lines (183 loc) · 5.24 KB

memory-checkpoint.md

File metadata and controls

220 lines (183 loc) · 5.24 KB
tools
read
write
bash
grep
arguments $CHECKPOINT_NAME

Memory Checkpoint Creation

Create a checkpoint of the current semantic tree state for recovery, branching, or milestone marking.

Based on the WFGY project: https://github.com/onestardao/WFGY

Instructions

  1. Prepare Checkpoint

    • Parse checkpoint name from "$CHECKPOINT_NAME"
    • If empty, auto-generate: "checkpoint_[timestamp]"
    • Read active tree from .wfgy/trees/active_tree.json
    • Calculate current state metrics

  2. Capture State Snapshot

    • Tree structure and all nodes
    • Current context from .wfgy/context.json
    • Active configuration
    • Boundary maps
    • Formula states:
      • BBMC residue levels
      • BBPF path states
      • BBAM attention distribution
      • Current ΔS and λ_observe

  3. Create Checkpoint Package

    {
  "id": "checkpoint_uuid",
  "name": "$CHECKPOINT_NAME",
  "timestamp": "ISO_8601",
  "tree_state": {
    "tree_name": "name",
    "node_count": N,
    "last_node": "node_id",
    "nodes": [...]
  },
  "context_state": {
    "current_deltaS": value,
    "lambda": "direction",
    "e_resonance": value,
    "active_module": "BBMC/BBPF/BBCR/BBAM"
  },
  "metrics": {
    "avg_deltaS": value,
    "stability": value,
    "confidence": percentage
  },
  "metadata": {
    "reason": "manual/auto/pre-risk",
    "tags": ["milestone", "stable", "experimental"]
  }
}

  4. Store Checkpoint

    • Save to .wfgy/checkpoints/[tree_name]/[checkpoint_name].json
    • Update checkpoint index
    • Create recovery instructions
    • Log in .wfgy/logs/checkpoints.log
    • Maintain max checkpoints (default: 50)

  5. Verify Checkpoint

    • Test checkpoint integrity
    • Validate all data preserved
    • Ensure recovery possible
    • Calculate checkpoint size
    • Update checkpoint registry

Output Format

CHECKPOINT CREATED SUCCESSFULLY
═══════════════════════════════════════
Checkpoint Name: $CHECKPOINT_NAME
Checkpoint ID: [uuid]
Created: [timestamp]

State Captured:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Tree: [tree_name]
Nodes: [count]
Position: Node [id] "[topic]"

Current Metrics:
• ΔS: [value]
• λ: [direction]
• E_resonance: [value]
• Confidence: [percentage]%

Checkpoint Statistics:
┌─────────────────────────────────────┐
│ Metric         │ Value  │ Status    │
├─────────────────────────────────────┤
│ Completeness   │ 100%   │ ✓ Full    │
│ Size           │ [size] │ Normal    │
│ Compression    │ [ratio]│ Applied   │
│ Recovery Time  │ ~[sec] │ Fast      │
└─────────────────────────────────────┘

Checkpoint Contents:
• Semantic nodes: [count]
• Relationships: [count]
• Context state: Preserved
• Configuration: Saved
• Boundaries: Mapped

Tags Applied:
[#manual] [#stable] [#milestone]

Recovery Instructions:
────────────────────────────────────────
To restore this checkpoint:
/memory:restore "$CHECKPOINT_NAME"

To branch from this checkpoint:
/semantic:tree-init "new_branch"
/memory:restore "$CHECKPOINT_NAME"

To compare with current:
/memory:diff "$CHECKPOINT_NAME"

Checkpoint Location:
.wfgy/checkpoints/[tree_name]/$CHECKPOINT_NAME.json

Related Checkpoints:
• Previous: [previous_checkpoint] ([time_ago])
• Next: None (latest)
• Similar: [similar_checkpoint]

Checkpoint Types

Manual Checkpoint

User-initiated save point

/memory:checkpoint "before_major_change"

Automatic Checkpoint

System-triggered saves:

  • Every N nodes (default: 50)
  • Before high-risk operations
  • At stability points
  • Daily backups

Recovery Checkpoint

Created before BBCR operations

/memory:checkpoint --type recovery

Milestone Checkpoint

Major achievement markers

/memory:checkpoint "v1.0_complete" --milestone

Checkpoint Management

List Checkpoints

/memory:checkpoint --list

Delete Old Checkpoints

/memory:checkpoint --cleanup --keep 10

Export Checkpoint

/memory:checkpoint --export "checkpoint_name"

Compare Checkpoints

/memory:checkpoint --compare "checkpoint1" "checkpoint2"

Auto-Checkpoint Rules

Configure in .wfgy/config.json:

{
  "auto_checkpoint": {
    "enabled": true,
    "frequency": 50,
    "on_risk": true,
    "risk_threshold": 0.7,
    "max_checkpoints": 50,
    "cleanup_age_days": 30
  }
}

Recovery Scenarios

  1. Logic Failure: Return to last stable
  2. Exploration: Branch for experiments
  3. Milestone: Mark achievements
  4. Backup: Regular safety saves
  5. Comparison: Track progress

Integration

Checkpoints work with:

  • /memory:restore to recover state
  • /wfgy:bbcr for failure recovery
  • /semantic:tree-switch for branches
  • /boundary:risk-assess for auto-saves