retention-cleanup

Retention Cleanup Skill

When to Use

Use this skill for artifact lifecycle management:

• Clean up expired screenshots/recordings/reports
• Enforce per-run size limits
• Free disk space
• Compliance with data retention policies

Instructions

Automatic Cleanup

Configured via retention config:

• rawVideoDays: 3 - Delete recordings older than 3 days
• screenshotsDays: 14 - Delete screenshots older than 14 days
• reportsDays: 90 - Delete reports older than 90 days
• maxRunSizeMb: 500 - Max size per run
• deleteExpiredAutomatically: true - Run on schedule

Manual Cleanup

1. Call delete_artifacts(paths[], confirm: true) for specific files
2. Or use retention manager directly for bulk operations

Run Size Enforcement

When adding artifacts to a run:

1. Check getRunSize(runId)
2. If > maxRunSizeMb, delete oldest artifacts in that run until under limit

Retention Policies by Type

Type	Default Retention	Use Case
Recordings	3 days	Large, transient, for debugging
Screenshots	14 days	Medium, reference for reports
Reports	90 days	Small, long-term reference

Safety

• Never delete outside approvedDirectories
• Always require confirm: true for manual deletion
• Log all deletions with timestamp, reason, file list
• Respect uploadArtifactsByDefault: false - don't auto-upload

Commands

# Automatic (runs on init and periodically)
retentionManager.cleanup()

# Manual specific files
delete_artifacts(["runs/screenshots/old.png", "runs/recordings/old.webm"], true)

# Check run size
retentionManager.getRunSize("run-123")

# Enforce limit
retentionManager.enforceRunSizeLimit("run-123")

Integration

• Called automatically after each run completes
• Hooked into capture_screenshot and record_interaction via event listeners
• Reports deleted artifacts in run summary

Examples

User: "Clean up old recordings from last week"
Agent:
1. retentionManager.cleanup() → returns count deleted
2. Report: "Deleted 47 recordings older than 3 days"