Skip to content

Releases: williamzujkowski/moltdown

v1.2.0 - Agent Resilience and Crash Recovery

02 Feb 18:35
@williamzujkowski williamzujkowski
v1.2.0
1b00ef1
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
v1.2.0 - Agent Resilience and Crash Recovery Latest
Latest

Overview

This release adds comprehensive agent process resilience features to prevent and recover from AI CLI crashes, particularly Claude Code memory leaks that can reach 13-120GB+ in extended sessions.

Key Features

🛡️ Agent Resilience Phase

New bootstrap phase 8 installs crash prevention and recovery tools:

Feature Description
Memory Watchdog systemd service that warns at 8GB, kills Claude at 13GB
cgroups Limiting run-claude-limited enforces hard memory limits
Session Persistence agent-session tmux wrapper with auto-reattach
Trend Analysis vm-health-check --trend predicts OOM 30-60 min ahead

📊 Resource Planning

  • Default RAM: 8GB → 16GB
  • Default swap: 4GB → 8GB
  • New --memory and --vcpus flags for setup_cloud.sh
  • New RESOURCES.md with parallel deployment guide

🧹 Cleanup

  • Removed .mcp.json from git (local config)
  • Removed vestigial guest/vm-health-check.sh
  • Fixed README accuracy

New Commands (Inside VM)

vm-health-check              # Quick health with Claude memory tracking
vm-health-check --watch      # Continuous monitoring (30s refresh)
vm-health-check --trend      # Memory trend analysis with OOM prediction
run-claude-limited           # Run Claude with 12GB hard limit
run-claude-limited 8G        # Custom limit
agent-session                # Persistent tmux session
systemctl status claude-watchdog  # Check watchdog service

Recommended 64GB Host Configuration

# Golden image (16GB, stopped during parallel work)
./setup_cloud.sh --memory 16384 --vcpus 4

# 3 parallel clones at 12GB each
./clone_manager.sh create ubuntu2404-agent --linked --memory 12288 --vcpus 4
./clone_manager.sh create ubuntu2404-agent --linked --memory 12288 --vcpus 4
./clone_manager.sh create ubuntu2404-agent --linked --memory 12288 --vcpus 4

Full Changelog

See CHANGELOG.md for complete details.

Full Changelog: v1.1.0...v1.2.0

Assets 2
Loading

moltdown v1.1.0 - Parallel Agent Workflows

02 Feb 01:33
@williamzujkowski williamzujkowski
v1.1.0
b510bca
This commit was signed with the committer’s verified signature.
williamzujkowski William Zujkowski
GPG key ID: 2B1A6D370DE1D7FF
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
moltdown v1.1.0 - Parallel Agent Workflows

What's New in v1.1.0

🚀 Parallel Agent Workflows

Create instant linked clones for running multiple agents simultaneously:

./clone_manager.sh create golden-vm --linked
./clone_manager.sh create golden-vm --linked
./clone_manager.sh create golden-vm --linked
# All 3 clones ready in seconds, fully isolated

☁️ Cloud Image Support

New recommended setup path using Ubuntu cloud images (~8 min vs ~20 min):

./setup_cloud.sh

🖥️ GUI Access

Connect to VMs graphically:

virt-viewer <vm-name>
# or use virt-manager for full management

🔑 SSH Key Authentication

Clones inherit SSH keys from golden image — no password prompts needed.

⏱️ Long-Running Sessions

VMs now configured for day/week-long agent runs:

  • 4GB swap file for memory pressure
  • Cloud-init disabled after bootstrap
  • Journal limits to prevent disk fill
  • Health monitoring: vm-health-check

🔒 Security

All VM disk images, credentials, and SSH keys stay local on your machine. Only scripts and documentation are in this repo. See Security Notes for details.

Performance

Operation Time
Create linked clone ~1 second
Clone boot to SSH ~150 seconds
Snapshot revert ~2 seconds

Full Changelog

See CHANGELOG.md for complete details.

Loading