# jonnyzzz Published articles for jonnyzzz. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Install, Launch, Update: Shipping a CLI on Five Platforms -- the devrig Approach DevFeed: [Install, Launch, Update: Shipping a CLI on Five Platforms -- the devrig Approach]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-08-06T00:00:00Z Content type: tutorial Language: en Sources: [Eugene Petrenko]() Topics: [Command-line interface](), [Kotlin](), [ide]() Tags: [agents](), [ai-agents](), [ai-coding](), [architecture](), [cli](), [command-line](), [devrig](), [distribution](), [eugene-petrenko](), [jdk](), [jonnyzzz](), [jvm](), [kotlin](), [mcp-steroid](), [open-source](), [sha-256](), [tag-41199d53f463]() ### AI overview This tutorial presents the devrig approach to distributing and updating a JVM command-line tool across five platforms. It describes generated install scripts with verified JDK coordinates and baked-in URLs and SHA-256 hashes, content-addressed side-by-side installations, binary-owned launchers, and updates performed by running a fresh install script. ### Source excerpt A JVM command-line tool is easy to write and miserable to ship. You need five platforms, a JDK the user does not have, an update path that never overwrites a running install, and a trust story for a `curl | sh` one-liner. Here is the architecture we ended up with in `devrig`: generate the install script from a model whose JDK coordinates are signature-verified, make installs content-addressed and side-by-side, let the binary own its own launcher, and make "update" mean nothing more than running a fresh install script ## Building a Personal Assistant Orchestrator with Claude Code and Headless Sub-Agents DevFeed: [Building a Personal Assistant Orchestrator with Claude Code and Headless Sub-Agents]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-07-19T00:00:00Z Content type: article Language: en Sources: [Eugene Petrenko]() Topics: [AI Agent](), [Claude Code](), [Code](), [Processes](), [Git](), [Slack](), [OpenClaw]() Tags: [ai-agent](), [ai-agents](), [ai-coding](), [architecture](), [claude-code](), [code](), [comparison](), [eugene-petrenko](), [git](), [hermes-agent](), [infrastructure](), [jonnyzzz](), [openclaw](), [process](), [prompt](), [self-improvement](), [slack](), [tag-41199d53f463](), [workflow]() ### AI overview The article describes building a personal-assistant orchestrator from a prompt in an interactive Claude Code session. It routes requests from Slack and Telegram, assigns each task to a fresh headless sub-agent process, and uses files, process boundaries, and git to manage work and recovery. The author also compares the approach with OpenClaw and Hermes. ### Source excerpt In one afternoon -- one pasted prompt, included below -- we turned an interactive Claude Code session into a personal-assistant orchestrator: channels in, a fresh headless sub-agent per request, an inbox loop that delivers finished work before taking new work, and a three-lens quorum that reviews and ships its own process fixes. By day two it was also babysitting the machine's other AI Agent consoles through a guarded keystroke path. No daemon, no webhooks, no database -- jsonl files, blocking CLIs, and git as the deployment mechanism -- and a hands-on comparison against OpenClaw and Hermes. ## A Real macOS Desktop for a Linux Agent: Driving Tart VMs over SSH with Agent Skills DevFeed: [A Real macOS Desktop for a Linux Agent: Driving Tart VMs over SSH with Agent Skills]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-07-16T00:00:00Z Content type: tutorial Language: en Sources: [Eugene Petrenko]() Topics: [Agent Skills](), [AI Agent](), [macOS](), [ssh](), [Linux](), [Command-line interface](), [Open Source]() Tags: [agent](), [agent-skills](), [ai-agent](), [ai-agents](), [ai-coding](), [cli](), [eugene-petrenko](), [intellij](), [jonnyzzz](), [linux](), [macos](), [open-source](), [ssh](), [tag-41199d53f463](), [tart](), [testing]() ### AI overview The article explains how a Linux AI Agent can connect over SSH to an Apple Silicon Mac, boot a Tart virtual machine, and capture and control a real macOS desktop. It presents reusable open-source Agent Skills and a CLI for moving this workflow across agents such as Claude Code, Codex, and Gemini. ### Source excerpt Last time we put a full IntelliJ IDEA inside a Docker container. But Docker only gives you Linux, and some GUIs are macOS -- Aqua, Gatekeeper, TCC prompts and all. So we taught a Linux AI Agent to boot, see, and drive a real macOS desktop inside a Tart VM over SSH, and packaged the whole thing as reusable Agent Skills you can point at any Apple Silicon Mac. ## A Real IntelliJ IDEA Inside Docker: Integration Tests with X Server, Window Manager, and Live Video DevFeed: [A Real IntelliJ IDEA Inside Docker: Integration Tests with X Server, Window Manager, and Live Video]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-07-05T00:00:00Z Content type: tutorial Language: en Sources: [Eugene Petrenko]() Topics: [IntelliJ IDEA](), [Docker](), [Testing](), [Docker Container](), [FFmpeg (Fast Forward Moving Picture Experts Group)](), [GUI](), [Node.js](), [MCP]() Tags: [ai-agents](), [ai-coding](), [browser](), [devrig](), [docker](), [docker-container](), [eugene-petrenko](), [ffmpeg](), [gui](), [intellij](), [intellij-idea](), [jonnyzzz](), [mcp](), [mcp-steroid](), [node-js](), [plugin](), [tag-41199d53f463](), [testing](), [tests]() ### AI overview This tutorial presents a reusable approach for running full IntelliJ IDEA integration tests inside Docker with a GUI. It covers the container image, X server, window management, IDE control, and streaming live video to a browser, with open-source code in the test-integration module. ### Source excerpt Evals that boot a full IntelliJ IDEA -- pixels, dialogs, indexing and all -- inside a Docker container. Xvfb provides the display, fluxbox keeps the windows in place, ffmpeg records everything, and a tiny Node.js server streams the live video straight into a browser tab on your machine. Here is the whole recipe, so you can build your own. ## Inside the Git Hooks: Tagging Every AI Agent Commit (Part 3) DevFeed: [Inside the Git Hooks: Tagging Every AI Agent Commit (Part 3)]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-06-26T00:00:00Z Content type: tutorial Language: en Sources: [Eugene Petrenko]() Topics: [Git](), [hooks](), [AI Agent](), [observability](), [Security]() Tags: [ai-agent](), [ai-agents](), [ai-coding](), [capture](), [commit](), [eugene-petrenko](), [git](), [git-hooks](), [infrastructure](), [jonnyzzz](), [provenance](), [repo](), [sandbox](), [tag-41199d53f463](), [workflow]() ### AI overview This third-part tutorial explains a best-effort Git hook layer for AI agents running inside a sandbox. The hooks stamp session IDs on commits, preserve the tags through amend, rebase, and squash, and capture restorable copies of pushed changes while continuing to run repository-local hooks. ### Source excerpt Part 3 -- the build for the lighter solution. How a set of git hooks stamps a session id on every commit an AI Agent makes, survives squash and rebase, and captures each push -- all best-effort, all inside the sandbox. ## Inside the Git Proxy: Capturing What an AI Agent Pushed (Part 2) DevFeed: [Inside the Git Proxy: Capturing What an AI Agent Pushed (Part 2)]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-06-22T00:00:00Z Content type: article Language: en Sources: [Eugene Petrenko]() Topics: [Git](), [AI Agent](), [HTTP](), [Authentication]() Tags: [ai-agent](), [ai-agents](), [ai-coding](), [authentication](), [eugene-petrenko](), [git](), [http](), [https](), [infrastructure](), [jonnyzzz](), [logs](), [proxy](), [refs](), [tag-41199d53f463](), [workflow]() ### AI overview This article explains the capture half of a Git proxy for AI agents. It covers intercepting HTTPS traffic, parsing Git packfiles, logging commits, pinning objects with marker refs, and handling signed and atomic protocol behavior. ### Source excerpt Part 2 -- how the git proxy intercepts HTTPS, forces a parseable pack, logs every commit, pins the objects with marker refs, and survives the sharp edges of a signed, atomic protocol. ## What Did the Agent Just Push? Auditing Git for AI Agents DevFeed: [What Did the Agent Just Push? Auditing Git for AI Agents]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-06-20T00:00:00Z Content type: tutorial Language: en Sources: [Eugene Petrenko]() Topics: [Git](), [AI Agent](), [audit](), [Clash](), [hooks](), [repo]() Tags: [ai-agents](), [ai-coding](), [audit](), [eugene-petrenko](), [git](), [hooks](), [infrastructure](), [jonnyzzz](), [proxy](), [pull-request](), [repo](), [tag-41199d53f463](), [workflow]() ### AI overview This article presents an approach for auditing autonomous AI agents that work with Git repositories. It recommends keeping real credentials away from agents, using a Git proxy as a hard boundary to limit which branches can change, and using Git hooks as a lighter, best-effort layer for provenance and telemetry. ### Source excerpt When you hand a repo to an autonomous AI Agent, you need to bound what it can touch and capture what it did. Part 1 -- the approach -- keep the keys away from the agent, a git proxy as the hard boundary, and git hooks as the lighter alternative. ## devrig: MCP Steroid Goes Level 3 -- Autonomous IDE Setup for AI Agents DevFeed: [devrig: MCP Steroid Goes Level 3 -- Autonomous IDE Setup for AI Agents]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-06-02T00:00:00Z Content type: article Language: en Sources: [Eugene Petrenko]() Topics: [Model Context Protocol (MCP)](), [AI Agent](), [IntelliJ IDEA](), [Command-line interface](), [Claude](), [Keycloak]() Tags: [agent](), [agentic-experience](), [ai-agent](), [ai-agents](), [ai-coding](), [claude](), [cli](), [devrig](), [eugene-petrenko](), [intellij](), [jonnyzzz](), [keycloak](), [mcp](), [mcp-steroid](), [tag-41199d53f463]() ### AI overview The article presents Level 3 of MCP Steroid, where the devrig CLI autonomously configures an agent's MCP connection and downloads or starts an IntelliJ-family IDE. Its end-to-end test reports that Claude found 4,633 PSI usages of a class in Keycloak in one prompt. ### Source excerpt The new `devrig` CLI in MCP Steroid 0.100 turns the IDE bring-up ritual into one stdio command. `devrig install ` writes the MCP config; `devrig backend download/start` brings an IntelliJ-family IDE into existence and gives the agent an IDE of its own. In the e2e test, Claude provisions an IDE and finds 4,633 PSI usages of one class in Keycloak -- in one prompt. ## From HTTP MCP to stdio: Lessons Connecting an AI Agent to a Desktop IDE DevFeed: [From HTTP MCP to stdio: Lessons Connecting an AI Agent to a Desktop IDE]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-05-20T00:00:00Z Content type: article Language: en Sources: [Eugene Petrenko]() Topics: [Model Context Protocol](), [IntelliJ IDEA](), [AI Agent](), [Command-line interface](), [Cross-origin resource sharing (CORS)]() Tags: [agentic-coding](), [ai-agent](), [ai-agents](), [ai-coding](), [architecture](), [claude](), [cli](), [codex](), [devrig](), [docker-image](), [eugene-petrenko](), [integration-test](), [intellij](), [jonnyzzz](), [kotlin](), [ktor](), [mcp](), [mcp-steroid](), [report](), [tag-41199d53f463]() ### AI overview This article explains why an HTTP MCP server embedded in a desktop IntelliJ application proved unreliable on real workstations. It describes moving to stdio with a CLI coordinator called devrig, and emphasizes integration tests for the MCP layer, including issues involving dynamic ports, startup ordering, multiple IDEs, and mismatched transport support. ### Source excerpt The obvious way to connect an AI Agent to a running IntelliJ is an HTTP MCP server inside the IDE. I built it, shipped it, and used it daily -- and it kept breaking on real workstations: dynamic ports, start-order lottery, multiple IDEs, and the agent-up/IDE-down dead end. Here is why I moved the whole thing to stdio and a CLI coordinator called devrig, and what the journey taught me. ## 130 Release-Roadmap Issues, 818 Agent Runs: A Mid-Iteration Reality Check DevFeed: [130 Release-Roadmap Issues, 818 Agent Runs: A Mid-Iteration Reality Check]() Original publisher: [Read original article]() Author: Eugene Petrenko Published: 2026-04-28T00:00:00Z Content type: article Language: en Sources: [Eugene Petrenko]() Topics: [AI Agent](), [Claude](), [codex](), [Artificial Intelligence](), [issue tracker](), [Orchestration](), [Command-line interface](), [Graphs](), [Requirements]() Tags: [agent](), [agents](), [ai](), [ai-agent](), [ai-agents](), [ai-coding](), [cli](), [code](), [eugene-petrenko](), [graph](), [issue-tracker](), [jonnyzzz](), [mcp](), [mcp-steroid](), [multi-agent](), [orchestration](), [pipeline](), [requirements](), [run-agent](), [tag-41199d53f463]() ### AI overview The article describes using a swarm of Claude, Codex, and Gemini agent CLIs to research 130+ release-roadmap issues and produce implementation specifications. The process used 818 agent runs over four days at an approximately $576 cost, followed by a mid-release comparison of the research with the team's actual work. ### Source excerpt We turned one release's roadmap -- 130+ umbrella-epic issues -- into fully-researched implementation specs using a swarm of three different agent CLIs (Claude, Codex, Gemini): 818 runs, ~155M+ tokens, ~$576, four days. Then, mid-release-cycle, we ran a snapshot to compare the research against reality -- with no expectation that anything was "done" yet. Here is the pipeline, the numbers, and the honest verdict.