Posts

If you SSH into servers, run long processes, or just want a better terminal experience, tmux changes everything. Sessions persist through disconnects, panes let you see multiple things at once, and once you build muscle memory, you’ll wonder how you worked without it. Why tmux? Sessions survive disconnects - SSH drops, laptop sleeps, terminal crashes? Your work continues. Split panes - Editor on the left, tests on the right, logs at the bottom. Multiple windows - Like browser tabs for your terminal. Pair programming - Share a session with someone else in real-time. Scriptable - Automate your development environment setup. Getting Started 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 # Start new session tmux # Start named session tmux new -s myproject # Detach (inside tmux) Ctrl+b d # List sessions tmux ls # Attach to session tmux attach -t myproject # Kill session tmux kill-session -t myproject The key prefix is Ctrl+b by default. Press it, release, then press the next key. ...

curl is the universal HTTP client. It’s installed everywhere, works with any API, and once mastered, becomes your go-to tool for testing, debugging, and scripting HTTP interactions. Basic Requests 1 2 3 4 5 6 7 8 # GET (default) curl https://api.example.com/users # Explicit methods curl -X POST https://api.example.com/users curl -X PUT https://api.example.com/users/1 curl -X DELETE https://api.example.com/users/1 curl -X PATCH https://api.example.com/users/1 Adding Headers 1 2 3 4 5 6 7 8 # Single header curl -H "Authorization: Bearer token123" https://api.example.com/me # Multiple headers curl -H "Authorization: Bearer token123" \ -H "Content-Type: application/json" \ -H "Accept: application/json" \ https://api.example.com/users Sending Data JSON Body 1 2 3 4 5 6 7 8 9 10 11 12 13 14 # Inline JSON curl -X POST https://api.example.com/users \ -H "Content-Type: application/json" \ -d '{"name": "Alice", "email": "alice@example.com"}' # From file curl -X POST https://api.example.com/users \ -H "Content-Type: application/json" \ -d @payload.json # From stdin echo '{"name": "Alice"}' | curl -X POST https://api.example.com/users \ -H "Content-Type: application/json" \ -d @- Form Data 1 2 3 4 5 6 7 8 # URL-encoded (default for -d without Content-Type) curl -X POST https://api.example.com/login \ -d "username=alice&password=secret" # Multipart form (file uploads) curl -X POST https://api.example.com/upload \ -F "file=@document.pdf" \ -F "description=My document" Response Handling Show Headers 1 2 3 4 5 6 7 8 # Response headers only curl -I https://api.example.com/health # Headers + body curl -i https://api.example.com/users # Verbose (request + response headers) curl -v https://api.example.com/users Output Control 1 2 3 4 5 6 7 8 9 10 11 12 13 14 # Save to file curl -o response.json https://api.example.com/users # Save with remote filename curl -O https://example.com/file.zip # Silent (no progress bar) curl -s https://api.example.com/users # Silent but show errors curl -sS https://api.example.com/users # Only output body (suppress all else) curl -s https://api.example.com/users | jq '.' Extract Specific Info 1 2 3 4 5 6 7 8 9 10 11 12 13 14 # HTTP status code only curl -s -o /dev/null -w "%{http_code}" https://api.example.com/health # Multiple variables curl -s -o /dev/null -w "Status: %{http_code}\nTime: %{time_total}s\nSize: %{size_download} bytes\n" \ https://api.example.com/users # Available variables # %{http_code} - HTTP status code # %{time_total} - Total time in seconds # %{time_connect} - Time to establish connection # %{time_starttransfer} - Time to first byte # %{size_download} - Downloaded bytes # %{url_effective} - Final URL after redirects Authentication 1 2 3 4 5 6 7 8 9 10 11 # Basic auth curl -u username:password https://api.example.com/secure # Bearer token curl -H "Authorization: Bearer eyJhbG..." https://api.example.com/me # API key in header curl -H "X-API-Key: abc123" https://api.example.com/data # API key in query string curl "https://api.example.com/data?api_key=abc123" Following Redirects 1 2 3 4 5 6 7 8 # Follow redirects (disabled by default) curl -L https://short.url/abc # Limit redirect count curl -L --max-redirs 5 https://example.com # Show redirect chain curl -L -v https://short.url/abc 2>&1 | grep "< location" Timeouts and Retries 1 2 3 4 5 6 7 8 9 10 11 # Connection timeout (seconds) curl --connect-timeout 5 https://api.example.com # Total operation timeout curl --max-time 30 https://api.example.com/slow-endpoint # Retry on failure curl --retry 3 --retry-delay 2 https://api.example.com # Retry on specific HTTP codes curl --retry 3 --retry-all-errors https://api.example.com SSL/TLS Options 1 2 3 4 5 6 7 8 # Skip certificate verification (development only!) curl -k https://self-signed.example.com # Use specific CA certificate curl --cacert /path/to/ca.crt https://api.example.com # Client certificate authentication curl --cert client.crt --key client.key https://api.example.com Cookies 1 2 3 4 5 6 7 8 9 10 11 # Send cookies curl -b "session=abc123; token=xyz" https://api.example.com # Save cookies to file curl -c cookies.txt https://api.example.com/login -d "user=alice&pass=secret" # Load cookies from file curl -b cookies.txt https://api.example.com/dashboard # Both (maintain session) curl -b cookies.txt -c cookies.txt https://api.example.com/action Useful Patterns Health Check Script 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 #!/bin/bash check_health() { local url=$1 local status=$(curl -s -o /dev/null -w "%{http_code}" --max-time 5 "$url") if [ "$status" = "200" ]; then echo "✓ $url" return 0 else echo "✗ $url (HTTP $status)" return 1 fi } check_health "https://api.example.com/health" check_health "https://web.example.com" API Testing 1 2 3 4 5 6 7 8 9 10 # Create resource and capture ID ID=$(curl -s -X POST https://api.example.com/users \ -H "Content-Type: application/json" \ -d '{"name": "Test User"}' | jq -r '.id') # Use captured ID curl -s https://api.example.com/users/$ID | jq '.' # Delete curl -X DELETE https://api.example.com/users/$ID Download with Progress 1 2 3 4 5 # Show progress bar curl -# -O https://example.com/large-file.zip # Resume interrupted download curl -C - -O https://example.com/large-file.zip Parallel Requests 1 2 3 4 5 6 7 8 # Using xargs echo -e "url1\nurl2\nurl3" | xargs -P 4 -I {} curl -s {} -o /dev/null -w "{}: %{http_code}\n" # Using curl's parallel feature (7.68+) curl --parallel --parallel-immediate \ https://api1.example.com \ https://api2.example.com \ https://api3.example.com Debugging Trace All Details 1 2 # Full trace including SSL handshake curl -v --trace-ascii debug.txt https://api.example.com Common Issues 1 2 3 4 5 6 7 8 9 # DNS resolution problems curl -v --resolve api.example.com:443:1.2.3.4 https://api.example.com # Force IPv4 or IPv6 curl -4 https://api.example.com # IPv4 only curl -6 https://api.example.com # IPv6 only # Use specific interface curl --interface eth0 https://api.example.com Config File Save common options in ~/.curlrc: ...

Environment files are deceptively simple. A few KEY=value pairs, what could go wrong? Quite a bit, actually. Here’s how to manage them without shooting yourself in the foot. The Basic Rules 1 2 3 4 # .env DATABASE_URL=postgres://localhost:5432/myapp API_KEY=sk_test_abc123 DEBUG=true Rule 1: Never commit secrets to git. 1 2 3 4 5 6 # .gitignore .env .env.local .env.*.local *.pem *.key Rule 2: Always commit an example file. 1 2 3 4 # .env.example (committed to repo) DATABASE_URL=postgres://localhost:5432/myapp API_KEY=your_api_key_here DEBUG=true New developers copy .env.example to .env and fill in their values. The example documents what’s needed without exposing real credentials. ...

Running AI workloads in containers presents unique challenges that traditional web application patterns don’t address. GPU scheduling, model caching, and bursty inference traffic all require thoughtful architecture. Here’s what actually works in production. The GPU Scheduling Problem Standard Kubernetes scheduling assumes CPU and memory are your primary constraints. When you add GPUs to the mix, everything changes. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 apiVersion: v1 kind: Pod metadata: name: inference-server spec: containers: - name: model image: my-registry/llm-server:v1.2 resources: limits: nvidia.com/gpu: 1 memory: "32Gi" requests: nvidia.com/gpu: 1 memory: "24Gi" The naive approach—one GPU per pod—works until you realize GPUs cost $2-4/hour and sit idle between requests. MIG (Multi-Instance GPU) and time-slicing help, but they introduce complexity: ...

If you work with APIs, logs, or configuration files, you work with JSON. And if you work with JSON from the command line, jq is indispensable. Here are the patterns I use daily. The Basics 1 2 3 4 5 6 7 8 9 10 # Pretty print echo '{"name":"alice","age":30}' | jq '.' # Extract a field echo '{"name":"alice","age":30}' | jq '.name' # "alice" # Raw output (no quotes) echo '{"name":"alice","age":30}' | jq -r '.name' # alice The -r flag is your friend. Use it whenever you want the actual value, not a JSON string. ...

SSH tunnels are one of those tools that seem magical until you understand the three basic patterns. Once you do, you’ll use them constantly. The Three Types Local forwarding (-L): Access a remote service as if it were local Remote forwarding (-R): Expose a local service to a remote network Dynamic forwarding (-D): Create a SOCKS proxy through the SSH connection Let’s break down each one. Local Forwarding: Reach Remote Services Locally Scenario: You need to access a database that’s only available from a server you can SSH into. ...

Makefiles have been around since 1976. They’re also still the best task runner for most projects. Here’s why, and how to use them effectively in 2026. The Case for Make Every ecosystem has its own task runner: npm scripts, Gradle, Rake, Poetry, Cargo. When your project spans multiple languages—a Python backend, TypeScript frontend, Go CLI tool, and Terraform infrastructure—you end up with five different ways to run tests. Make provides one interface: ...

Every developer knows the pain: you’re deep in a feature branch, someone asks you to review a PR, and now you’re stashing changes, switching branches, rebuilding dependencies, and losing your mental context. Git worktrees solve this elegantly. The Problem with Branch Switching Traditional git workflow assumes one working directory: 1 2 3 4 5 6 7 8 # You're working on feature-x git stash git checkout main git pull git checkout pr-branch npm install # dependencies changed npm run build # wait for it... # review, then reverse the whole process Every switch costs time and mental energy. Your editor loses state. Your dev server restarts. Your flow is gone. ...

Git hooks run scripts at key points in your workflow. Use them to catch problems before they become pull requests. Hook Basics Hooks live in .git/hooks/. Each hook is an executable script named after the event. 1 2 3 4 5 6 # List available hooks ls .git/hooks/*.sample # Make a hook active cp .git/hooks/pre-commit.sample .git/hooks/pre-commit chmod +x .git/hooks/pre-commit Common Hooks Hook When Use Case pre-commit Before commit created Lint, format, test prepare-commit-msg After default message Add ticket numbers commit-msg After message entered Validate format pre-push Before push Run full tests post-merge After merge Install dependencies post-checkout After checkout Environment setup Pre-commit Hook Basic Linting 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 #!/bin/bash # .git/hooks/pre-commit echo "Running pre-commit checks..." # Get staged files STAGED_FILES=$(git diff --cached --name-only --diff-filter=ACM) # Python files PYTHON_FILES=$(echo "$STAGED_FILES" | grep '\.py$') if [ -n "$PYTHON_FILES" ]; then echo "Linting Python files..." ruff check $PYTHON_FILES || exit 1 black --check $PYTHON_FILES || exit 1 fi # JavaScript files JS_FILES=$(echo "$STAGED_FILES" | grep -E '\.(js|ts)$') if [ -n "$JS_FILES" ]; then echo "Linting JavaScript files..." eslint $JS_FILES || exit 1 fi echo "All checks passed!" Prevent Debug Statements 1 2 3 4 5 6 7 8 9 #!/bin/bash # .git/hooks/pre-commit # Check for debug statements if git diff --cached | grep -E '(console\.log|debugger|import pdb|breakpoint\(\))'; then echo "ERROR: Debug statements found!" echo "Remove console.log, debugger, pdb, or breakpoint() before committing." exit 1 fi Check for Secrets 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 #!/bin/bash # .git/hooks/pre-commit # Patterns that might be secrets PATTERNS=( 'password\s*=\s*["\047][^"\047]+' 'api_key\s*=\s*["\047][^"\047]+' 'secret\s*=\s*["\047][^"\047]+' 'AWS_SECRET_ACCESS_KEY' 'PRIVATE KEY' ) for pattern in "${PATTERNS[@]}"; do if git diff --cached | grep -iE "$pattern"; then echo "ERROR: Possible secret detected!" echo "Pattern: $pattern" exit 1 fi done Commit Message Hook Enforce Conventional Commits 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 #!/bin/bash # .git/hooks/commit-msg COMMIT_MSG_FILE=$1 COMMIT_MSG=$(cat "$COMMIT_MSG_FILE") # Conventional commit pattern PATTERN="^(feat|fix|docs|style|refactor|test|chore)(\(.+\))?: .{1,50}" if ! echo "$COMMIT_MSG" | grep -qE "$PATTERN"; then echo "ERROR: Invalid commit message format!" echo "" echo "Expected: <type>(<scope>): <description>" echo "Types: feat, fix, docs, style, refactor, test, chore" echo "" echo "Examples:" echo " feat(auth): add OAuth2 support" echo " fix: resolve memory leak in worker" echo "" exit 1 fi Add Ticket Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 #!/bin/bash # .git/hooks/prepare-commit-msg COMMIT_MSG_FILE=$1 BRANCH_NAME=$(git symbolic-ref --short HEAD) # Extract ticket number from branch (e.g., feature/JIRA-123-description) TICKET=$(echo "$BRANCH_NAME" | grep -oE '[A-Z]+-[0-9]+') if [ -n "$TICKET" ]; then # Prepend ticket number if not already present if ! grep -q "$TICKET" "$COMMIT_MSG_FILE"; then sed -i.bak "1s/^/[$TICKET] /" "$COMMIT_MSG_FILE" fi fi Pre-push Hook Run Tests Before Push 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 #!/bin/bash # .git/hooks/pre-push echo "Running tests before push..." # Run test suite npm test if [ $? -ne 0 ]; then echo "ERROR: Tests failed! Push aborted." exit 1 fi # Check coverage threshold npm run coverage COVERAGE=$(cat coverage/coverage-summary.json | jq '.total.lines.pct') if (( $(echo "$COVERAGE < 80" | bc -l) )); then echo "ERROR: Coverage below 80%! Current: $COVERAGE%" exit 1 fi echo "All checks passed. Pushing..." Prevent Force Push to Main 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 #!/bin/bash # .git/hooks/pre-push PROTECTED_BRANCHES="main master" CURRENT_BRANCH=$(git symbolic-ref --short HEAD) # Check if force pushing while read local_ref local_sha remote_ref remote_sha; do if echo "$PROTECTED_BRANCHES" | grep -qw "$CURRENT_BRANCH"; then if [ "$remote_sha" != "0000000000000000000000000000000000000000" ]; then # Check if this is a force push if ! git merge-base --is-ancestor "$remote_sha" "$local_sha" 2>/dev/null; then echo "ERROR: Force push to $CURRENT_BRANCH is not allowed!" exit 1 fi fi fi done Post-merge Hook Auto-install Dependencies 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 #!/bin/bash # .git/hooks/post-merge CHANGED_FILES=$(git diff-tree -r --name-only ORIG_HEAD HEAD) # Check if package.json changed if echo "$CHANGED_FILES" | grep -q "package.json"; then echo "package.json changed, running npm install..." npm install fi # Check if requirements.txt changed if echo "$CHANGED_FILES" | grep -q "requirements.txt"; then echo "requirements.txt changed, running pip install..." pip install -r requirements.txt fi # Check if migrations added if echo "$CHANGED_FILES" | grep -q "migrations/"; then echo "New migrations detected!" echo "Run: python manage.py migrate" fi Using pre-commit Framework The pre-commit framework manages hooks declaratively. ...

YAML looks simple until it isn’t. These gotchas have broken production configs and wasted countless debugging hours. Learn them once, avoid them forever. The Norway Problem 1 2 3 4 5 # These are ALL booleans in YAML 1.1 country: NO # false answer: yes # true enabled: on # true disabled: off # false Fix: Always quote strings that could be interpreted as booleans. 1 2 country: "NO" answer: "yes" YAML 1.2 fixed this, but many parsers (including PyYAML by default) still use 1.1 rules. ...