Portable Claude Code workflow toolkit: skills, config, repo bootstrapper, and MCP bridge for multi-agent communication
This repository also configures its own agents. See what agentic-workflow tells them →
joi-fairshare/agentic-workflow
Agent Claude Code
Implements a single, well-scoped code change from an already-approved brief (goal, acceptance criteria, proof command). Use for one bounded edit — not for open-ended exploration, planning, or anything needing web search or MCP data tools.
joi-fairshare/agentic-workflow
Agent Claude Code
Read-only investigation — finds where code lives, answers "how does X work", summarizes a diff or a spec. Never edits files. Use instead of general-purpose or Explore when the answer needs to be synthesized (not just located) but the task needs no write access.
joi-fairshare/agentic-workflow
Agent Claude Code
Reviews a diff or PR against this repo's AGENTS.md and best-practices conventions. Report-only, never edits. Use for a quick pre-push check, not a full multi-agent /pr-review deep pass.
joi-fairshare/agentic-workflow
Agent Claude Code
Plans a short Playwright script (route, role, steps, expected state) against the local web-app stack from a diff and the route map. Does not execute browser automation itself — it writes the script for a no-model runner to execute. Use only for the planning half of UI evidence generation.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: