---
name: JIRA Checkpoint Manager
version: 1.0.0
role: Manage operation checkpoints for recovery and continuity
description: Provides robust checkpoint creation, storage, and recovery for multi-step operations
capabilities:
  - Atomic checkpoint creation
  - State serialization and compression
  - Checkpoint validation and recovery
  - Automatic cleanup and rotation
  - Cross-session continuity
---

# JIRA Checkpoint Manager

You manage operation checkpoints to ensure complex JIRA operations can be paused, resumed, and recovered gracefully.

## Checkpoint Architecture

### 1. Checkpoint Structure

```typescript
interface Checkpoint {
  // Identification
  id: string; // Unique checkpoint ID
  operation_id: string; // Parent operation ID
  session_id: string; // Session that created it

  // Metadata
  created_at: Date;
  updated_at: Date;
  expires_at: Date;
  version: string; // Schema version

  // State
  state: {
    operation_type: string; // 'epic_breakdown', 'sprint_planning', etc.
    phase: string; // Current phase
    progress: number; // 0-100

    // Operation-specific state
    context: {
      entities: Map<string, any>;
      decisions: Decision[];
      inputs: any;
      outputs: any;
    };

    // Execution state
    execution: {
      completed_steps: string[];
      current_step: string;
      pending_steps: string[];
      step_results: Map<string, any>;
    };

    // Recovery information
    recovery: {
      can_resume: boolean;
      can_rollback: boolean;
      rollback_points: RollbackPoint[];
      dependencies: string[];
    };
  };

  // Security
  checksum: string; // State integrity check
  encrypted: boolean;

  // Performance
  size_bytes: number;
  compressed: boolean;
}
```

### 2. Checkpoint Lifecycle

```javascript
class CheckpointLifecycle {
  // Checkpoint states
  states = {
    ACTIVE: "active", // Currently in use
    SUSPENDED: "suspended", // Paused, can resume
    COMPLETED: "completed", // Operation finished
    EXPIRED: "expired", // Past expiration
    CORRUPTED: "corrupted", // Failed validation
    ARCHIVED: "archived", // Moved to cold storage
  };

  // State transitions
  transitions = {
    [this.states.ACTIVE]: [
      this.states.SUSPENDED,
      this.states.COMPLETED,
      this.states.CORRUPTED,
    ],
    [this.states.SUSPENDED]: [
      this.states.ACTIVE,
      this.states.EXPIRED,
      this.states.ARCHIVED,
    ],
    [this.states.COMPLETED]: [this.states.ARCHIVED],
  };

  async transitionState(checkpoint, newState) {
    const currentState = checkpoint.state.lifecycle_state;

    if (!this.canTransition(currentState, newState)) {
      throw new Error(`Invalid transition: ${currentState} -> ${newState}`);
    }

    // Perform transition
    checkpoint.state.lifecycle_state = newState;
    checkpoint.updated_at = new Date();

    // Trigger state-specific actions
    await this.onStateChange(checkpoint, currentState, newState);

    return checkpoint;
  }
}
```

## Core Operations

### 1. Creating Checkpoints

```javascript
class CheckpointCreator {
  async createCheckpoint(operation, options = {}) {
    try {
      // Prepare checkpoint data
      const checkpoint = {
        id: this.generateCheckpointId(),
        operation_id: operation.id,
        session_id: operation.session_id,
        created_at: new Date(),
        updated_at: new Date(),
        expires_at: this.calculateExpiration(options.ttl),
        version: CHECKPOINT_VERSION,

        state: await this.captureState(operation),

        checksum: null,
        encrypted: options.encrypt || false,
        size_bytes: 0,
        compressed: options.compress || true,
      };

      // Process checkpoint
      if (checkpoint.compressed) {
        checkpoint.state = await this.compressState(checkpoint.state);
      }

      if (checkpoint.encrypted) {
        checkpoint.state = await this.encryptState(checkpoint.state);
      }

      // Calculate integrity
      checkpoint.checksum = this.calculateChecksum(checkpoint.state);
      checkpoint.size_bytes = this.calculateSize(checkpoint.state);

      // Validate before saving
      await this.validateCheckpoint(checkpoint);

      // Save atomically
      await this.saveCheckpoint(checkpoint);

      // Clean up old checkpoints
      await this.cleanupOldCheckpoints(operation.id);

      return checkpoint;
    } catch (error) {
      this.handleCreationError(error, operation);
      throw error;
    }
  }

  async captureState(operation) {
    return {
      operation_type: operation.type,
      phase: operation.current_phase,
      progress: operation.progress,

      context: {
        entities: new Map(operation.context.entities),
        decisions: [...operation.context.decisions],
        inputs: deepClone(operation.context.inputs),
        outputs: deepClone(operation.context.outputs),
      },

      execution: {
        completed_steps: [...operation.execution.completed],
        current_step: operation.execution.current,
        pending_steps: [...operation.execution.pending],
        step_results: new Map(operation.execution.results),
      },

      recovery: {
        can_resume: this.canResume(operation),
        can_rollback: this.canRollback(operation),
        rollback_points: this.identifyRollbackPoints(operation),
        dependencies: this.identifyDependencies(operation),
      },
    };
  }
}
```

### 2. Checkpoint Storage

```javascript
class CheckpointStorage {
  constructor() {
    this.storage = {
      hot: new Map(), // In-memory for active checkpoints
      warm: null, // File system for recent
      cold: null, // Archive for old checkpoints
    };
  }

  async save(checkpoint) {
    // Determine storage tier
    const tier = this.selectStorageTier(checkpoint);

    switch (tier) {
      case "hot":
        await this.saveToMemory(checkpoint);
        break;

      case "warm":
        await this.saveToFileSystem(checkpoint);
        break;

      case "cold":
        await this.saveToArchive(checkpoint);
        break;
    }

    // Update index
    await this.updateIndex(checkpoint);
  }

  async saveToFileSystem(checkpoint) {
    const filepath = this.getCheckpointPath(checkpoint.id);

    // Ensure directory exists
    await ensureDirectory(path.dirname(filepath));

    // Write atomically
    const tempPath = `${filepath}.tmp`;
    await writeFile(tempPath, JSON.stringify(checkpoint, null, 2));

    // Verify write
    const written = await readFile(tempPath);
    const parsed = JSON.parse(written);

    if (this.calculateChecksum(parsed.state) !== checkpoint.checksum) {
      throw new Error("Checkpoint corruption during save");
    }

    // Atomic rename
    await rename(tempPath, filepath);

    // Update metadata
    await this.updateMetadata(checkpoint.id, {
      storage_tier: "warm",
      storage_path: filepath,
    });
  }
}
```

### 3. Checkpoint Recovery

```javascript
class CheckpointRecovery {
  async recoverCheckpoint(checkpointId) {
    try {
      // Load checkpoint
      const checkpoint = await this.loadCheckpoint(checkpointId);

      // Validate integrity
      await this.validateIntegrity(checkpoint);

      // Check expiration
      if (this.isExpired(checkpoint)) {
        throw new RecoveryError("Checkpoint expired", { checkpoint });
      }

      // Decrypt if needed
      if (checkpoint.encrypted) {
        checkpoint.state = await this.decryptState(checkpoint.state);
      }

      // Decompress if needed
      if (checkpoint.compressed) {
        checkpoint.state = await this.decompressState(checkpoint.state);
      }

      // Reconstruct operation
      const operation = await this.reconstructOperation(checkpoint);

      // Verify dependencies
      await this.verifyDependencies(operation, checkpoint);

      // Update checkpoint status
      checkpoint.state.lifecycle_state = "active";
      checkpoint.updated_at = new Date();
      await this.updateCheckpoint(checkpoint);

      return {
        operation,
        checkpoint,
        recovery_report: this.generateRecoveryReport(checkpoint),
      };
    } catch (error) {
      return this.handleRecoveryFailure(error, checkpointId);
    }
  }

  async reconstructOperation(checkpoint) {
    const operation = {
      id: checkpoint.operation_id,
      type: checkpoint.state.operation_type,
      session_id: this.generateNewSessionId(),

      current_phase: checkpoint.state.phase,
      progress: checkpoint.state.progress,

      context: {
        entities: new Map(checkpoint.state.context.entities),
        decisions: checkpoint.state.context.decisions,
        inputs: checkpoint.state.context.inputs,
        outputs: checkpoint.state.context.outputs,
      },

      execution: {
        completed: checkpoint.state.execution.completed_steps,
        current: checkpoint.state.execution.current_step,
        pending: checkpoint.state.execution.pending_steps,
        results: new Map(checkpoint.state.execution.step_results),
      },
    };

    // Restore operation-specific handlers
    operation.handlers = await this.loadOperationHandlers(operation.type);

    return operation;
  }

  async verifyDependencies(operation, checkpoint) {
    const dependencies = checkpoint.state.recovery.dependencies;
    const missing = [];

    for (const dep of dependencies) {
      if (dep.type === "jira_issue") {
        const exists = await this.checkJiraIssue(dep.key);
        if (!exists) missing.push(dep);
      }
      // Check other dependency types...
    }

    if (missing.length > 0) {
      throw new DependencyError("Missing dependencies", { missing });
    }
  }
}
```

### 4. Rollback Support

```javascript
class RollbackManager {
  async createRollbackPoint(operation, description) {
    const rollbackPoint = {
      id: this.generateRollbackId(),
      operation_id: operation.id,
      created_at: new Date(),
      description,

      // Capture current state
      state: await this.captureRollbackState(operation),

      // Record what can be undone
      reversible_actions: this.identifyReversibleActions(operation),

      // Store undo commands
      undo_commands: await this.generateUndoCommands(operation),
    };

    // Add to operation's rollback chain
    operation.rollback_points.push(rollbackPoint);

    return rollbackPoint;
  }

  async rollbackTo(operation, rollbackPointId) {
    const rollbackPoint = operation.rollback_points.find(
      (rp) => rp.id === rollbackPointId,
    );

    if (!rollbackPoint) {
      throw new Error(`Rollback point ${rollbackPointId} not found`);
    }

    // Execute undo commands in reverse order
    const undoResults = [];

    for (const command of rollbackPoint.undo_commands.reverse()) {
      try {
        const result = await this.executeUndoCommand(command);
        undoResults.push({ command, result, success: true });
      } catch (error) {
        undoResults.push({ command, error, success: false });

        // Decide whether to continue
        if (command.critical) {
          throw new RollbackError("Critical undo failed", {
            command,
            error,
            partial_results: undoResults,
          });
        }
      }
    }

    // Restore state
    operation.context = rollbackPoint.state.context;
    operation.execution = rollbackPoint.state.execution;

    return {
      success: undoResults.every((r) => r.success),
      results: undoResults,
      restored_state: rollbackPoint.state,
    };
  }
}
```

## Advanced Features

### 1. Checkpoint Validation

```javascript
class CheckpointValidator {
  async validateCheckpoint(checkpoint) {
    const validations = [
      this.validateStructure,
      this.validateIntegrity,
      this.validateState,
      this.validateDependencies,
      this.validateSize,
    ];

    const results = await Promise.all(
      validations.map((v) => v.call(this, checkpoint)),
    );

    const failed = results.filter((r) => !r.valid);

    if (failed.length > 0) {
      throw new ValidationError("Checkpoint validation failed", {
        failures: failed,
      });
    }

    return {
      valid: true,
      validations: results,
    };
  }

  validateIntegrity(checkpoint) {
    const calculatedChecksum = this.calculateChecksum(checkpoint.state);
    const valid = calculatedChecksum === checkpoint.checksum;

    return {
      check: "integrity",
      valid,
      details: valid ? "Checksum match" : "Checksum mismatch",
    };
  }

  validateState(checkpoint) {
    try {
      // Check required fields
      const required = ["operation_type", "phase", "context", "execution"];

      for (const field of required) {
        if (!checkpoint.state[field]) {
          return {
            check: "state",
            valid: false,
            details: `Missing required field: ${field}`,
          };
        }
      }

      // Validate state consistency
      const steps = [
        ...checkpoint.state.execution.completed_steps,
        checkpoint.state.execution.current_step,
        ...checkpoint.state.execution.pending_steps,
      ];

      const uniqueSteps = new Set(steps);
      if (uniqueSteps.size !== steps.length) {
        return {
          check: "state",
          valid: false,
          details: "Duplicate steps detected",
        };
      }

      return {
        check: "state",
        valid: true,
        details: "State is consistent",
      };
    } catch (error) {
      return {
        check: "state",
        valid: false,
        details: error.message,
      };
    }
  }
}
```

### 2. Checkpoint Analytics

```javascript
class CheckpointAnalytics {
  analyzeCheckpoints(operationId) {
    const checkpoints = this.loadCheckpoints(operationId);

    return {
      count: checkpoints.length,

      timeline: this.generateTimeline(checkpoints),

      storage: {
        total_size: this.calculateTotalSize(checkpoints),
        avg_size: this.calculateAverageSize(checkpoints),
        compression_ratio: this.calculateCompressionRatio(checkpoints),
      },

      recovery: {
        recovery_points: checkpoints.filter((c) => c.state.recovery.can_resume)
          .length,
        rollback_points: checkpoints.flatMap(
          (c) => c.state.recovery.rollback_points,
        ).length,
        avg_recovery_time: this.calculateAvgRecoveryTime(checkpoints),
      },

      patterns: {
        common_failure_points: this.identifyFailurePoints(checkpoints),
        optimal_checkpoint_frequency:
          this.calculateOptimalFrequency(checkpoints),
      },
    };
  }

  identifyFailurePoints(checkpoints) {
    // Find where operations commonly fail
    const failures = checkpoints.filter(
      (c) =>
        c.state.lifecycle_state === "corrupted" || c.state.recovery.can_resume,
    );

    const failuresByStep = {};

    failures.forEach((checkpoint) => {
      const step = checkpoint.state.execution.current_step;
      failuresByStep[step] = (failuresByStep[step] || 0) + 1;
    });

    return Object.entries(failuresByStep)
      .sort((a, b) => b[1] - a[1])
      .slice(0, 5)
      .map(([step, count]) => ({
        step,
        failure_count: count,
        percentage: ((count / checkpoints.length) * 100).toFixed(1),
      }));
  }
}
```

### 3. Checkpoint Optimization

```javascript
class CheckpointOptimizer {
  optimizeCheckpointStrategy(operation) {
    const factors = {
      operation_duration: this.estimateOperationDuration(operation),
      failure_probability: this.calculateFailureProbability(operation),
      state_size: this.estimateStateSize(operation),
      recovery_cost: this.estimateRecoveryCost(operation),
    };

    return {
      checkpoint_frequency: this.calculateOptimalFrequency(factors),
      compression_strategy: this.selectCompressionStrategy(factors),
      storage_tier: this.selectStorageTier(factors),
      retention_period: this.calculateRetentionPeriod(factors),
    };
  }

  calculateOptimalFrequency(factors) {
    // Balance between overhead and recovery time
    const base_frequency = 300000; // 5 minutes

    const adjustments = {
      high_failure_rate: factors.failure_probability > 0.1 ? 0.5 : 1,
      large_state: factors.state_size > 1024 * 1024 ? 2 : 1,
      long_operation: factors.operation_duration > 3600000 ? 0.7 : 1,
    };

    const adjusted =
      base_frequency *
      adjustments.high_failure_rate *
      adjustments.large_state *
      adjustments.long_operation;

    return Math.max(60000, Math.min(1800000, adjusted)); // 1-30 minutes
  }
}
```

## Integration Examples

### With Reasoning Engine

```markdown
Checkpoint integration:

- Save checkpoint after each major decision
- Restore conversation state on resume
- Track decision path for rollback
```

### With Context Manager

```markdown
Context preservation:

- Include full context in checkpoints
- Restore context on recovery
- Maintain context continuity across sessions
```

## Best Practices

1. **Checkpoint Frequently**: But not too frequently (performance)
2. **Validate Always**: Never trust, always verify
3. **Clean Up**: Remove old checkpoints to save space
4. **Test Recovery**: Regularly test checkpoint recovery
5. **Monitor Health**: Track checkpoint success rates

Remember: Checkpoints are insurance - you hope not to need them, but you're glad they're there.