​

Best Practices

​

Security Recommendations

​

Command Approval Strategy

​

Auto-Approve Patterns

kubectl get
kubectl describe
kubectl logs
git status
git log
git diff
ls
cat
pwd
docker ps

kubectl delete
kubectl apply
rm
git push
terraform apply
terraform destroy
npm publish

​

Integrated Environment Workflows

​

Chat + Terminal + Files

1. Chat (right): “The API deployment is failing in staging”
2. Terminal (center bottom): See logs appear
   Copy

   $ kubectl logs deployment/api -n staging
   Error: ConfigMap 'api-config' not found
   
3. File Browser (left): Navigate to kubernetes/staging/api/
4. Editor (center top): Open deployment.yaml, spot the typo
5. Edit the file: Fix api-confg → api-config
6. Chat: “Apply the fixed deployment”
7. Terminal: Watch the apply succeed

​

Context Switching Best Practices

• Cmd/Ctrl+` - Toggle Terminal visibility
• Cmd/Ctrl+/ - Focus Chat input
• Click file in tree - Open in editor
• Click Terminal/Activity tabs - Switch views

• Keep relevant files open in tabs
• Use Terminal tab for active execution
• Switch to Activity tab to review history
• Minimize/maximize panels as needed

​

Using Workspaces Effectively

/Users/you/projects/frontend
/Users/you/projects/backend
/Users/you/infra/k8s-configs

/Users/you  # Too broad

/etc
/var
/usr/local

​

File Editing Strategy

• Quick config tweaks
• Viewing files during chat
• Small changes suggested by Calmo
• Files you want to reference in conversation

• Large refactorings
• Complex multi-file changes
• Debugging sessions
• Long coding sessions

​

Workflow Optimization

​

Effective Prompting

• “Check the logs for the api-server pod in the staging namespace, last 50 lines”
• “Show me all pods in CrashLoopBackOff state across all namespaces”
• “List recent commits in this repo with their authors”

• “Check the logs” (which pod? which namespace?)
• “What’s wrong?” (too vague for targeted commands)

​

Batch Operations

"First, check the pod status in production.
Then, describe any pods that aren't running.
Finally, show me the logs from failing containers."

• Approve them one by one as you review
• Click Allow on all if they’re safe
• Deny any that look incorrect

​

Leveraging MCP Servers

​

Development Workflows

​

Kubernetes Debugging

​

Infrastructure Changes

1. Edit in Bridge - Make changes to YAML/HCL files
2. Preview first - Run terraform plan or kubectl diff (safe, auto-approve-able)
3. Review changes - Read the plan output in Terminal
4. Apply intentionally - Manually approve terraform apply or kubectl apply
5. Verify - Check resources with get commands

You: [Edit deployment.yaml to increase replicas]
You: "Apply this deployment to staging"
Calmo: kubectl apply -f deployment.yaml -n staging
[Command Approval panel shows MEDIUM RISK]
You: [Review, click Allow]
Terminal: deployment.apps/api configured
You: "How many replicas are running now?"
Calmo: kubectl get deployment api -n staging
Terminal: api  3/3  3  3  10s

​

Git Operations

• git status, git log, git diff
• git branch -a, git remote -v
• git show <commit>

• git commit - Review commit message
• git push - Ensure correct branch
• git merge - Understand what’s being merged
• git checkout - Verify branch/file changes

1. Chat: “Show me the status and recent changes”
2. Review output in Terminal
3. Chat: “Commit these changes with message: fix: correct API endpoint”
4. Review commit command, approve
5. Chat: “Push to feature branch”
6. Verify branch name, approve

​

Performance Tips

​

Keep Bridge Running

• Close window → Minimizes to menu bar/tray
• Connection persists → No re-pairing needed
• History preserved → Command and chat history available
• MCP servers stay connected → No reload time

• After updating Bridge version
• If memory usage is very high
• When troubleshooting connection issues

​

Manage Pending Commands

• Approve or deny promptly - Review and decide
• Set up auto-approve for safe patterns
• Clear queue if needed - Deny commands you don’t need

​

Regular Cleanup

• Click Clear in Activity tab when log grows large
• Doesn’t affect command history with Calmo
• Frees up memory

• Close file tabs you’re no longer referencing
• Cmd/Ctrl+W to close active tab
• Right-click → Close All to clear all tabs

​

Team Usage

​

Individual Setup

• Each person pairs their own machine - One Bridge per developer
• Personal approval patterns - Auto-approve lists are per-user
• Individual workspaces - Add your own project directories

​

Shared Standards

# Team Auto-Approve Patterns

Safe for Always:
- kubectl get
- kubectl describe
- kubectl logs
- git status
- git log

Always Manual:
- kubectl delete
- kubectl apply
- terraform apply
- git push

{
  "mcpServers": {
    "context7": {
      "url": "https://mcp.context7.com/mcp",
      "headers": {
        "CONTEXT7_API_KEY": "${YOUR_KEY_HERE}"
      }
    },
    "sequential-thinking": {
      "command": "/usr/local/bin/npx",
      "args": ["-y", "@modelcontextprotocol/server-sequential-thinking"]
    }
  }
}

• Production systems → High scrutiny, minimal auto-approve
• Staging/dev → More auto-approve, faster iteration
• Local only → Maximum freedom

​

What to Avoid

• ❌ Editing production configs without review
• ❌ Running commands blindly without reading them
• ❌ Adding workspaces you don’t actively use
• ❌ Keeping 50+ file tabs open
• ❌ Ignoring risk level indicators

​

Quick Reference

​

Risk Assessment Matrix

​

Workspace Scope Guidelines

​

MCP Server Selection Guide

​

Troubleshooting Your Workflow

​

Commands Taking Too Long

1. Check command scope - Are you querying too much data?
   • Use -n namespace instead of -A (all namespaces)
   • Add --tail=100 to log commands
   • Use | head -n 20 to limit output
2. Network latency - Remote clusters take longer
   • Use kubectl contexts for faster access
   • Consider running commands on the cluster directly
3. Optimize patterns - Make commands more specific

​

Losing Context in Chat

1. Reference open files - “Look at the deployment.yaml I have open”
2. Be explicit - “In the staging namespace” instead of “there”
3. Use workspace context - Calmo sees your workspace structure
4. Provide recent terminal output - “Based on the logs you just showed me…”

​

Terminal Output Hard to Read

1. Use filtering - | grep, | less, | head
2. Request formatted output - Ask Calmo to summarize
3. Resize panels - Drag dividers for more space
4. Copy to editor - Copy output, paste in new file for analysis