#claude-code
14 articles with this tag.
Telegram bot as a remote control for the agent — architecture and safety
Telegram is my main interface to Claude Code when I'm away from the terminal. Showing how I set up the bot, secured it against prompt injection, and how it talks to the agent.
Prompt caching — how I calculate Opus 4.7 costs (and why the bill isn't scary)
Opus 4.7 looks expensive. Five minutes of prompt caching cuts cost by 90%. Showing how it works, how to measure, and which cadence patterns actually move the bill.
Memory in Claude Code — what it actually remembers and how to steer it
Memory is a Claude Code mechanism that persists facts about you, the project and preferences across sessions. Showing how it works under the hood, when to use it, and when CLAUDE.md is better.
Writing your own Claude Code skills — zero to MVP in 30 minutes
A skill is a small markdown file that adds a slash command with dedicated logic to Claude Code. Showing how to write one from scratch, where to put it, and which mistakes to avoid.
Background agents and Ralph loop — when they pay off, when they burn tokens
Background agents and Ralph-style loops are powerful, but easy to turn into a token furnace. Showing my three real cases and the rule for when to even turn this on.
MCP servers I actually use — review after 6 months
I have 12 MCP servers wired in — Home Assistant, Proxmox, Cloudflare, Vikunja, Gmail, Calendar, n8n. Showing which actually pay off and which sat unused until I unhooked them.
Plan mode in Claude Code — when to turn it on, when not to bother
Plan mode shipped recently and already changed my workflow. When I use it, when it slows me down, and what questions to ask so the plan is actually useful.
claudeclaw + cron — Claude as an assistant with its own schedule
Claude Code can work even when I'm not driving. claudeclaw is the mechanism that schedules tasks for it — daily briefings, PR reviews, dependency audits. Showing how it works on my setup.
Hooks in Claude Code — how I enforce discipline on the agent
Hooks are a small piece of config that turns Claude Code into a tool with real guardrails. Showing my setups: blocking destructive commands, enforcing commit policy, telemetry to Telegram.
Claude Opus 4.7 — first impressions after a week with 1M context
Opus 4.7 ships with a million token context, plan mode and a fast toggle. I switched from Sonnet 4.6 as my default — here's what changed, where it pays off, and where I stick with the cheaper model.
Claude Code skills in 2026: what's worth knowing, what to avoid, and why 36% try to rob you
Skills ate slash commands, the plugin marketplace has 1,234 entries, official Anthropic skills are doing design work, and ToxicSkills found prompt injection in over a third of them. Practical ecosystem map: what to install, what to scan, what to delete.
MCP in 2026 — from hype to grown-up tech with grown-up problems
A year ago Model Context Protocol was supposed to change everything. Today it's a standard, but the ecosystem has 20k servers, 92% exploitation odds at 10 plugins, and context bloat eating 70% of the window. Market overview, top 15, and 2026 trends.
How Claude Code changed my workflow — automations that run 24/7
A concrete tour of how an AI agent works on my server: from Telegram through Home Assistant to deploying sites. What works, what doesn't, and how to set it up.
Spec-Driven Development in Claude Code — why I no longer write 'just build me X' prompts
A five-stage process for planning a feature before the first line of code. Fewer iterations, fewer bugs, less refactoring. Concrete example: the very blog you're reading.