| name | logvalet:draft |
| description | Draft a context-aware Backlog issue comment: progress update, inquiry, resolution notice, or escalation — the LLM generates a draft based on the issue's full context and comment history. TRIGGER when: user says "コメント下書き", "draft comment", "コメントを書いて", "返信を作って", "コメントを作成", "comment draft", "コメント草稿", "コメントを下書きして", "issue comment を書いて", "バックログにコメント", "進捗報告コメント", "確認依頼コメント", "解決通知コメント", "コメントで報告したい", "返信案を作って", "エスカレーションコメント", "write a comment", "reply to this issue", "コメントの文面を考えて". DO NOT TRIGGER when: user wants to create a new issue (use issue-create) or wants to update issue fields like status/assignee (use logvalet CLI directly). Workflow: Automatically fetches context via /logvalet:context before drafting.
|
logvalet-draft
lv issue context を材料に、LLM がコンテキストに沿ったコメント下書きを生成する。
For full logvalet CLI documentation, see the logvalet skill.
When to use this skill
Use logvalet-draft when you need to:
- write a progress update comment for an in-progress issue
- draft a confirmation request to a stakeholder
- compose a resolution notice when closing an issue
- reply to a question in the issue comment thread
- write any issue comment where context (history, tone, status) matters
Workflow
Step 1: Identify issue key
If the user provides a Backlog issue key (e.g., PROJ-123, a URL like *.backlog.com/view/PROJ-123), extract and use it directly.
If no key is provided, ask for it.
Step 2: Clarify comment purpose
Ask the user (if not already clear) what kind of comment they want to write:
- 進捗報告 (progress update)
- 確認依頼 (confirmation request)
- 解決通知 (resolution notice)
- 質問応答 (reply to a question)
- その他 (other — let the user describe)
This helps the LLM calibrate tone and content.
Step 3: Fetch issue context
lv issue context ISSUE_KEY -f json
This returns:
- issue details (summary, description, status, priority, assignee, dates)
- comment history (tone, participants, recent discussion)
- analysis signals (stale indicator, overdue flag, blocker hints)
Step 4: Draft the comment
Using the issue context, generate a comment that:
- matches the tone of past comments in the thread (formal/casual, Japanese/English)
- reflects the current issue state (e.g., stale → acknowledge delay; overdue → address deadline)
- fits the stated purpose (progress update, inquiry, etc.)
- is concise and actionable
Step 5: Present draft for review
## コメント下書き — ISSUE_KEY
> コンテキスト: 最終更新 N日前、担当者: UserName、ステータス: X
---
<下書き内容>
---
このコメントを投稿しますか?[はい/編集/キャンセル]
If the user wants to edit, revise and re-present.
Step 6: Post the comment
After user confirmation, write the draft to a temp file and post:
lv issue comment add ISSUE_KEY --content-file /tmp/draft_ISSUE_KEY.md --dry-run
lv issue comment add ISSUE_KEY --content-file /tmp/draft_ISSUE_KEY.md
Always run --dry-run first before posting.
Notes
issue context provides comment history — use it to match the thread's existing tone and avoid repeating information already stated
- If the issue has no previous comments, default to a polite, professional Japanese tone
- The draft should reference specific issue details (e.g., deadline, assignee name) to feel contextual, not generic
- For long drafts, write to a temp file (
/tmp/draft_ISSUE_KEY.md) rather than passing via --content
Anti-patterns
- Do not post a comment without showing the draft and confirming with the user first
- Do not use a generic template without incorporating issue-specific context
- Do not guess the issue key — always confirm
- Do not skip
--dry-run before posting