Randoneering Agent Guide

Skills, agents, hooks, commands, and integration docs for OpenCode, pi, and other coding agents.

Skills

Category	Skills
Core Workflow	using-superpowers, brainstorming, writing-plans, executing-plans
Development Workflow	test-driven-development, subagent-driven-development, dispatching-parallel-agents, using-git-worktrees, finishing-a-development-branch
Quality	systematic-debugging, damage-control, verification-before-completion
Code Review	requesting-code-review, receiving-code-review
Writing	documentation, writing-skills, writing-style, resume-tailor
Data	dbt, postgres, snowflake, clickhouse, neon
Development Stacks	python, nix, automation, flox
Cloud & Infrastructure	cloudflare, hashicorp
Security	trail_of_bits, sentry

Agents

Category	Contents
Engineering	task-focused specialists for implementation, architecture, security, DevOps, and review
Marketing	content, campaign, paid media, SEO, and growth agents
Product	research, prioritization, feedback synthesis, and product strategy agents
Project Management	planning, tracking, workflow, and delivery-oriented agents
Specialized	domain-specific specialists across sales, finance, legal, and other functions
Spatial Computing	XR, AR/VR, and spatial interface agents
Academic	research, writing, and analysis agents for academic work
Testing	evidence, API, accessibility, performance, and validation agents
Support	reporting, compliance, infrastructure, finance, and support agents
Strategy	orchestration docs, playbooks, runbooks, and coordination guides
Integrations	converted formats and install docs for supported agent tools

Structure

agents/
├── {category}/
│   └── <domain>-<role>.md   # Agent definition
│
skills/
└── {category}/
    ├── {skill}/
    │   ├── SKILL.md       # Skill definition
    │   └── references/    # Supporting docs (optional)
    └── README.md          # Category or package notes (optional)

Usage

• OpenCode: copy skills into .claude/skills/ or load them from a shared repo path referenced by AGENTS.md
• pi: copy skills into .pi/skills/ or .agents/skills/, or point pi at this repo via its skills setting
• Agent files in agents/ are plain Markdown prompts with portable name and description frontmatter; use them as prompt templates, imported agents, or source material for tool-specific packaging

Configuration

• OpenCode reads project instructions from AGENTS.md
• pi reads AGENTS.md and also supports CLAUDE.md
• Use repo-local AGENTS.md for shared instructions when targeting both harnesses

Compatibility

• Skills use the Agent Skills SKILL.md structure, which both OpenCode and pi can consume
• Skills may include optional frontmatter such as allowed-tools, compatibility, metadata, or user-invocable; pi ignores unknown fields and OpenCode can still use the instructions
• Agent files keep only the portable frontmatter subset: name and description
• Runtime features still vary by harness. Skills that mention subagents, todo tools, or platform-specific commands may require harness-specific adaptation during execution

License

GPLv3 - see LICENSE