Claude Code guide in three levels, Beginner to Advanced: hands-on lessons with checked exercises, every volatile fact cited to its source. Independent community guide.
Autoprompt is a coding-agent skill that cuts failures by 45% on agentic coding tasks.
A DeepSeek coding agent harness: persistent shell, Ultra subagents, auto approval, Chrome MCP and session telemetry
Development workflows for Claude Code that keep broad exploration focused on the outcome you approved.
Kether-governed AI agent orchestration with Pi/MCP, isolated workers, LSP tools, and a portable Windows/WSL2 installer
See what your coding agent actually did. A desktop pal and notch for Claude Code, Codex, Cursor, Gemini CLI and any agent: test results read from the output, faked passes caught, and an MCP server any AI tool can ask. New: talk to Blu, who sees your screen, draws on it as it explains and uses your apps.
One Claude Code session, two subscriptions: Claude orchestrator + GPT executor subagents via a credential-clean loopback splitter
A live tree of the Codex and Claude Code sessions, subagents and teammates working in a repository
Delete DSH conversations instead of archiving them, see the whole family a conversation spawned (its subagents and derived conversations), and delete in bulk. / 真正删除 DeepSeek Harness 里的会话,并看清它生出的子智能体与派生对话。
The Claude Code harness I run every day, published under this name since day one and now the same repository as ucsandman/Agnostic-AI: guard hooks, rules, context modules, Mods, subagents, workflows, tools and jobs, ported to 20 clients. Zero-dependency Node.
One command turns every agent session already on your disk into a graph: subagents, tool calls, retries, the moments you said no. Ask your agent what happened and the graph lights up the nodes it means. npx rungraph: no hooks, no setup, all local. Works with Claude Code, Codex, Hermes, opencode and Cursor.