coding-agents

Close your MacBook and let your AI coding agents keep working. A small macOS menu-bar app for Apple silicon that stops the lid from suspending long-running work.

Working practices for shipping production software with coding agents — extracted from running a live multi-tenant B2B SaaS solo.

Open-source AI coding cheat sheets for Claude Code, Codex, Cursor, Copilot, Gemini CLI, MCP, AGENTS.md, CLAUDE.md, prompts, workflows, and team security.

File-level isolation for AI coding agents: run parallel implementation lanes without git worktrees.

Local-first agent operating system for durable AI coding: task graphs, deterministic context packs, executable delegation, verification evidence, and approval gates.

Self-hosted federated runtime for observable coding-agent organizations: signed graph nodes, distributed runners, Nostr coordination, and git-backed artifacts.

Native desktop app for local coding agents — plus on-device wallets and X Layer deploy. Rust, GPUI, no cloud account.

Cross-platform Python library and CLI for bounded subprocess execution with timeouts, output limits, process-tree cleanup, and health checks.

A local-first taskboard and CLI for coding-agent workflows.

Archived experiment: approval-gated workspace runtime for AI coding agents.

Shared local memory and coordination for concurrent AI coding agents

Explain and safely stop dev servers left running by coding agents.

State-driven, verifiable, low-context delivery workflow for AI coding agents.

Local-first agent-forensics workbench for Codex and Claude Code sessions: timeline, transcript, diff, insights, and redacted reports.

AI-created POC CLI for routing coding tasks across multiple AI coding agents.

AI Agent Orchestrator 2026 – Turn GitHub Issues Into Autonomous Workflows

Choose the minimum sufficient execution tactic for coding-agent work: direct by default, with focused discovery and fresh verification only when justified.

Zero-round-trip human routing for pre-generated coding-agent command bundles

Your agent writes the code. Humanoid writes the history. A shadow git remote that rewrites AI-generated commit history into atomic, verified, reviewable pull requests.

Local verification layer for Claude Code and Codex — lifecycle hooks that track whether changed code is backed by a passing check of exactly that code, and nudge the agent to run the smallest relevant test before finishing.