Portable snapshot of a pi coding-agent setup + backup/restore tooling. Clone to rebuild 22 extensions on another machine. Excludes secrets, cache, and machine state.
-
Updated
Sep 20, 2026 - Shell
Portable snapshot of a pi coding-agent setup + backup/restore tooling. Clone to rebuild 22 extensions on another machine. Excludes secrets, cache, and machine state.
The best Pi coding agent setup — 10 packages, 59 skills, 10-step autonomous workflow. Zero bloat, pure Pi ideology.
Muapi-hosted coding-model benchmark and setup guides for Codex, Claude Code, and OpenCode; compare coding quality and refusal behavior.
龙翼Hermes 自进化方法论 · 让 AI Agent 从"陌生外包"变成"懂你的合伙人"——Hermes 一键初始化,新老用户零配置增量升级、零副作用,Agent 越用越懂你
Disciplined code review, OWASP-style security checks, safe refactoring, and dependency and secret hygiene, with findings ranked by severity and backed by evidence.
Research with cited sources, deep-read papers into Markdown or Obsidian-style vaults, keep a living wiki, and hand work between sessions without losing context or project memory.
Full-stack TypeScript with Next.js App Router and Supabase: Postgres schema and RLS, auth, Tailwind UI, end-to-end tests and a careful release routine.
SEO and content for developer products and startups: landing page copy, technical SEO, launch planning and social posts, written honestly and without scraping or paid APIs.
Automatically generates SOUL.md and STYLE.md templates based on agent purpose to eliminate 30-60 minutes of manual setup work
For people building their own agent setup: write skills, AGENTS.md and CLAUDE.md files and sub-agents, manage context, and audit instructions for drift.
Planning-first workflow for coding agents: turn a rough idea into a spec, record architecture decisions, and break the work into small verifiable tasks before any code is written.
A setup for building Flutter, iOS, Android and React Native apps: platform conventions, simulator testing, and an app-store readiness checklist before you ship.
Test-first development, reliable unit and end-to-end tests, evidence-based root-cause debugging, and real-browser verification, for any language or framework.
Hermes Agent配置工作流:config设置/API key/fallback链
Typed Python services and data work: FastAPI-style backends, pytest, SQL and Postgres design and migrations, data quality checks and tidy notebooks.
Plan-first infrastructure work: Terraform, AWS, containers, CI/CD pipelines, safe deployment and observability, with nothing applied or deployed until you approve it.
Write READMEs, technical docs, changelogs, PR descriptions and ADRs with an agent, strip AI-sounding prose, and draw diagrams that earn their place in the docs.
Build frontends with real design taste and measurable quality: UI/UX direction, accessibility, Core Web Vitals, purposeful animation and design-system tokens, checked in a real browser.
Edit, grade, mix and export DevRel and social videos in DaVinci Resolve with an agent: cut long screen recordings, clean podcast audio, add captions and lower-thirds, and render every platform's version.
To associate your repository with the agent-setup topic, visit your repo's landing page and select "manage topics."