ワンクリックでManusで任意のスキルを実行

opencode-build-subagent

スター0

フォーク0

更新日2026年5月14日 06:08

Explicit-only opencode delegation skill for research drafts, source discovery, timeline/IOC drafts, file/text edits, code exploration, first-pass triage, and review-assist candidate discovery. Codex must verify primary evidence before final conclusions.

インストール

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

Manusで実行

ソース

zereight

zereight/skill

GitHub リポジトリを開く Creator のリポジトリを見る

ダウンロード

Manusで実行

opencode-build-subagent

Use this skill ONLY when the user explicitly asks for opencode delegation. Trigger keywords: '$opencode-build-subagent', '-build-subagent', 'use opencode', 'with opencode', 'opencode', 'opencode skill', or similar explicit opencode wording (including romanized Korean: opencode sseo, opencode ro). Do NOT auto-delegate without explicit user request.

Delegated task coverage

Security research drafts / candidate collection
Research drafts: source discovery, timeline/IOC/explanation drafts, comparison drafts
File/text edits, narrow code exploration, impact-candidate collection
First-pass test/log failure triage
Refactor/design drafts, docs/PR description drafts
Review-assist candidate discovery

Contract

Always call /Users/tao.exe/.codex/bin/opencode-build-subagent for delegated work.
The worker always runs opencode run with agent build.
Default model is opencode-go/deepseek-v4-flash.
The worker may edit the current checkout directly.
opencode output is candidate/helper output, NOT final evidence.
Codex must inspect primary evidence (file contents, git diff, test output, logs) before asserting findings or verification.
Keep work local only when the task is tiny, security-sensitive, destructive, blocked by permissions, or explicitly requires Codex-native tools.
Do not delegate: connector evidence reads, destructive decisions, final verification/conclusions after primary evidence, simple Q&A, current mode/status checks: these require Codex with primary evidence.

Required Follow-up

After every worker run:

Inspect git diff --stat and changed file names.
Review changed files against primary sources before asserting correctness.
Run the smallest relevant verification available.
Mention that opencode edited the checkout if files changed.
Do not treat opencode output as verified: re-check with primary evidence.

Example

/Users/tao.exe/.codex/bin/opencode-build-subagent \
  --dir /Users/tao.exe/Documents/bankx/mobile-app-workspace \
  "Implement the requested change. Keep edits minimal."

このリポジトリの他の Skills

同じリポジトリ

autocontext

zereight/skill

Run iterative agent evaluation and improvement loops (judge, improve, scenarios, playbooks) via pi-autocontext or autoctx CLI. Use when improving repeatable agent workflows (review rubrics, skill quality, scenario-based feedback), checking run status, or accumulating knowledge—not for one-off code edits or repo pattern drift (use continuity) or single PR review (use zereight-review).

2026-05-260

harness

zereight/skill

Pi-native evaluate→improve→persist harness. Provides a systematic loop for generating candidates, judging them against rubrics, and accumulating knowledge as playbooks. Use for iterative output improvement, not for one-off edits (use continuity) or single PR review (use zereight-review).

2026-05-260

skill-cleaner

zereight/skill

Audit agent skills — token cost, duplicates, outdated plugin versions, unused skills, and overly long descriptions. Use when trimming skill prompt budget, finding duplicate or unused skills, auditing plugin versions, or deciding which skills to remove.

2026-05-260

skill-not-showing

zereight/skill

Diagnose why a skill is missing from Pi /skill or Cursor lists. Use when a skill exists on disk but does not appear, Pi shows only project or extension skills, or after adding a repo skill under .agents/skills/.

2026-05-260

zereight-review

zereight/skill

Comprehensive code review skill for practical PR feedback. Use for feature, bugfix, and refactor reviews. Prioritizes correctness, edge cases, logic invariants, fallback-chain safety, async state transitions, architecture analysis, OWASP security, and clear actionable feedback.

2026-05-260

ljg-skill-map

zereight/skill

Skill map and inventory workflow. Use for scanning installed skills, skill overview, 스킬 목록, skill map.

2026-05-260

name	opencode-build-subagent
description	Explicit-only opencode delegation skill for research drafts, source discovery, timeline/IOC drafts, file/text edits, code exploration, first-pass triage, and review-assist candidate discovery. Codex must verify primary evidence before final conclusions.

opencode-build-subagent

Delegated task coverage

Security research drafts / candidate collection
Research drafts: source discovery, timeline/IOC/explanation drafts, comparison drafts
File/text edits, narrow code exploration, impact-candidate collection
First-pass test/log failure triage
Refactor/design drafts, docs/PR description drafts
Review-assist candidate discovery

Contract

Always call /Users/tao.exe/.codex/bin/opencode-build-subagent for delegated work.
The worker always runs opencode run with agent build.
Default model is opencode-go/deepseek-v4-flash.
The worker may edit the current checkout directly.
opencode output is candidate/helper output, NOT final evidence.
Codex must inspect primary evidence (file contents, git diff, test output, logs) before asserting findings or verification.
Keep work local only when the task is tiny, security-sensitive, destructive, blocked by permissions, or explicitly requires Codex-native tools.
Do not delegate: connector evidence reads, destructive decisions, final verification/conclusions after primary evidence, simple Q&A, current mode/status checks: these require Codex with primary evidence.

Required Follow-up

After every worker run:

Inspect git diff --stat and changed file names.
Review changed files against primary sources before asserting correctness.
Run the smallest relevant verification available.
Mention that opencode edited the checkout if files changed.
Do not treat opencode output as verified: re-check with primary evidence.

Example

/Users/tao.exe/.codex/bin/opencode-build-subagent \
  --dir /Users/tao.exe/Documents/bankx/mobile-app-workspace \
  "Implement the requested change. Keep edits minimal."