Skip to content
Agent Orchestration
Skill

/hoyeon-execute

Hoyeon execution workflow for Codex. Use when the user invokes "$hoyeon-execute" or wants to execute a Hoyeon plan.json through the Bash-first Codex adapter. This adapter loads the canonical execute skill and follows its Codex runtime surface.

BOOST
From plugin
hoyeon
17337 skills29 agents7 hooks
Install
$ npx -y skills add team-attention/hoyeon --skill hoyeon-execute --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/hoyeon-execute

Context preview

The summary Claude sees to decide when to auto-load this skill.

Hoyeon execution workflow for Codex. Use when the user invokes "$hoyeon-execute" or wants to execute a Hoyeon plan.json through the Bash-first Codex adapter. This adapter loads the canonical execute skill and follows its Codex runtime surface.

SKILL.md

hoyeon-execute.SKILL.md
name: hoyeon-execute
description: |
  Hoyeon execution workflow for Codex. Use when the user invokes
  "$hoyeon-execute" or wants to execute a Hoyeon plan.json through the
  Bash-first Codex adapter. This adapter loads the canonical execute skill and
  follows its Codex runtime surface.

hoyeon-execute

This is the Codex-facing wrapper for Hoyeon's canonical `execute` skill.

Canonical skill:

  • Installed root: `__HOYEON_PLUGIN_ROOT__/skills/execute/SKILL.md`
  • Repo-local fallback: `skills/execute/SKILL.md` from the current Hoyeon repo

When this skill is invoked:

1. Read the canonical skill file above before executing the workflow. 2. Follow the `Runtime Surface` -> `Codex` section in that file. 3. Mutate task state only through `hoyeon-cli plan task`. 4. Do not edit `plan.json` directly. 5. Do not rely on hooks, MCP, or Claude `TeamCreate` for Codex v1. 6. Use Hoyeon native-agent adapter names when dispatching subagents:

  • `worker` -> `hoyeon-worker`
  • `verifier` -> `hoyeon-verifier`
  • `code-reviewer` -> `hoyeon-code-reviewer`

7. In Codex, dispatch those adapters with the native `spawn_agent` tool:

  • `spawn_agent(agent_type="hoyeon-worker", message="<worker charter>")`
  • `spawn_agent(agent_type="hoyeon-verifier", message="<verification charter>")`
  • `spawn_agent(agent_type="hoyeon-code-reviewer", message="<review charter>")`

8. Treat canonical `Agent(...)`, `TaskCreate`, `TaskUpdate`, `TaskOutput`, and `TeamCreate` examples as Claude Code protocol notes, not literal Codex calls. 9. Parallel worker dispatch is allowed for disjoint `parallel_safe` tasks when the relevant adapter is prompt-visible. `scripts/codex-execute-smoke.sh` validates plan state transitions only, not parallel subagent behavior.

The output contract remains the canonical Hoyeon contract: all executable tasks completed or blocked with evidence, followed by plan validation.

Read more
Ships withhoyeon

All you need is requirements. A Claude Code plugin that derives requirements from your intent, verifies every derivation, and delivers traced code — without you writing a plan. AI can build anything. The hard part is knowing what to build — precisely.

Get the whole plugin

Other skills on hoyeon.