← ClaudeAtlas

unioss-knowledge-reportlisted

How to write a valuable ticket report from ticket evidence — structure, method, depth modes, user's language. Read before writing any today/ticket/daily report.
ttncode/unioss-plugins · ★ 0 · Code & Development · score 70
Install: claude install-skill ttncode/unioss-plugins
# UNIOSS Knowledge — Ticket Report The single definition of a valuable ticket report. Both digest flows read this before writing. ## Method - Read the ticket's FULL description and ALL notes from the evidence file — never summarize from the title alone. - Write the report in the **user's current conversation language** — never a fixed language. Translate Japanese source content into it. - Derive acceptance criteria from concrete statements in the ticket — never invent requirements. - Mark uncertainty explicitly in Open questions rather than guessing. - Never drop a ticket from a multi-ticket digest. - Report with the user's language, never use fixed language. ## Depth modes - **daily** (today / refresh daily): ticket-content depth — no codebase reading; omit Suggested direction or keep it ticket-content-level. - **single-ticket** (`/unioss-knowledge-ticket`): full depth — read the codebase before Suggested direction. Module path comes from `.walkthrough/.config/unioss.config.json` → `source.modules.admin-page` for AP tickets, `source.modules.front-end` for FE. Locate the screens/controllers/models the ticket touches and ground the direction in what exists. ## Report structure (one section per ticket) ```markdown ## AP#<iid> — <one-line title translated to the user's language> **Summary** — one sentence: what this ticket is. | | | |---|---| | **Who/When** | <author> · created <date> · state <opened/closed> | | **Ticket** | <web_url> | **What** - synthesize the actual