Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
Skyhook avatar

Skyhook Skills

  • Updated July 12, 2026
  • skyhook-io/skyhook-skills

skyhook-skills is a Claude Code plugin that provides composable slash-command loops for planning, implementation, review, and PR convergence so developers can run disciplined multi-agent workflows.

About

skyhook-skills is a Claude Code plugin packing composable dev-workflow slash commands, including full /autodev autopilot and focused plan, review, and PR loops. It adds cross-model review with Codex or Cursor and a product-review pass that questions what to build before judging code correctness. Use it when you want structured agent ceremonies that scale to task size instead of ad-hoc prompts for every feature.

  • /autodev runs plan → implement → review → PR → converge
  • Cross-model review via Codex or Cursor with skeptical triage
  • /product-review scores user journeys and UX comprehension
  • Composable loops inherit fixes from shared triage commands

Skyhook Skills by the numbers

  • Data as of Jul 13, 2026 (Skillselion catalog sync)
/plugin marketplace add skyhook-io/skyhook-skills

Add your badge

Show developers this marketplace is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Last updatedJuly 12, 2026
Repositoryskyhook-io/skyhook-skills

How do you orchestrate plan-implement-review-PR cycles in Claude Code without rebuilding prompts for every task?

Run /autodev, /plan-loop, and /review-loop commands to plan, implement, cross-review, and open PRs with Claude and Codex.

Who is it for?

Developers using Claude Code with optional Codex or Cursor who want gated loops and cross-model review on nontrivial changes.

Skip if: One-line fixes where full autodev ceremony would add more process than value.

What you get

Plans, triaged review findings, fixes, and PR updates until loops converge or hit configured caps.

Recommended Marketplaces

FAQ

Does cross-review auto-accept findings?

No—every finding from any reviewer is triaged with evidence; cross-model output is never auto-accepted.

How do I pick Codex versus Cursor as reviewer?

Set reviewer in ~/.claude/Skyhook Skills.json, SKYHOOK_REVIEWER, or inline consult directives on review commands.

Automationgittestingintegrations

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.