jira-reader
/jira-reader
Czyta zgłoszenie Jira wraz z komentarzami, podzadaniami i stronami Confluence, a potem zapisuje skrócony plik zadania w repozytorium bez zmian w Jira i Confluence.
- Autor
- editorgit ↗
- Repozytorium
- agentic-delivery ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 03.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 3
- Skrypty
- 2
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 03.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
- 2 skrypty w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.
Skrypty: scripts/fetch_issue.py, scripts/jira.py
Co robi i kiedy
Czyta zgłoszenie Jira wraz z komentarzami, podzadaniami i stronami Confluence, a potem zapisuje skrócony plik zadania w repozytorium bez zmian w Jira i Confluence.
Oryginalny opis (EN)
Read a Jira issue — description, comments, subtasks, links, attachments and the Confluence pages it links to — and write the distilled task file `.tracker/<KEY>.md` at the git root. Use when the user names a Jira key and wants it prepared for implementation or review, or pastes a Confluence page URL to read. The purpose is ALWAYS to end with the file written. READ-ONLY against Jira and Confluence.
Jak działa
- Workflow (every step is mandatory)
- Rules
Podgląd SKILL.md
--- name: jira-reader description: Read a Jira issue — description, comments, subtasks, links, attachments and the Confluence pages it links to — and write the distilled task file `.tracker/<KEY>.md` at the git root. Use when the user names a Jira key and wants it prepared for implementation or review, or pastes a Confluence page URL to read. The purpose is ALWAYS to end with the file written. READ-ONLY against Jira and Confluence. --- # jira-reader — Jira issue → `.tracker/<KEY>.md` The goal is a file with the task: the spec as Jira states it, close to verbatim, plus the essence and acceptance criteria. Reading the issue is the means; **writing the file is the required outcome**. Only Jira and Confluence are read-only. Credentials: `JIRA_BASE_URL`, `JIRA_EMAIL`, `JIRA_API_TOKEN` in the environment or in `~/.config/agentic-delivery/.env`. Never echo the token. ## Workflow (every step is mandatory) 1. **Fetch.** From this skill's base directory: `python3 scripts/fetch_issue.py <KEY>` It prints the issue, its comments, subtasks and links, downloads attachments to `.tracker/<KEY>/`, and expands the Confluence pages linked from the description, comments and remote links (first 5; `DELIVERY_CONFLUENCE_MAX=N` for more). A page with no text of its own lists its children: read the relevant one with `python3 scripts/fetch_issue.py --page <id>`. A bare Confluence URL from the user is answered the same way, with no task file. If the output has a `Parent:` line, fetch the parent too (one level up only). A subtask often holds just the bug or the rework, while the parent holds the original spec, its pages and its attachments. Look at downloaded images and PDFs (Read tool): a mockup is part of the spec. 2. **Quote the spec, synthesize the rest.** What the issue, comments and linked pages require goes into «Spec from Jira» close to verbatim. The debate around it is not dumped: what it settled is quoted as a requirement, what it left open becomes a question. 3. **Collect open questions** where comments conflict or scope and criteria are unclear. 4. **Confirm with the user** (when running directly, not inside the `ticket-reader` agent): show the essence, the criteria and the questions; their answers go to «Decided in discussion», the spec stays as Jira has it. 5. **Write the file** at `<git root>/.tracker/<KEY>.md` (`git rev-parse --show-toplevel`, never relative to the current directory). Create `.tracker/` if needed. Format: ```markdown # <KEY>: <one-line title> ## Essence 2–4 sentences: what to do and why. ## Spec from Jira (do not edit) Written only from Jira, by this skill or the ticket-reader agent. Nobody edits it afterwards — not the implementer, not the review loop: the reviewer checks the code against it. ### Issue text ### Requirements set in comments Every comment that sets, changes or cancels a requirement: author, date, text. ### From Confluence and attachments The parts that set requirements; the rest is a link under «Links». Omit if none. ### From the parent issue <PARENT-KEY> Only when there is a parent: the spec this subtask relies on, by the same rules. The scope is the subtask's own text; the parent is context to stay consistent with. Where they contradict, both stay and the conflict goes to «Open questions». ### Acceptance criteria - [ ] a checkable criterion that follows from the text above and only from it ## Decided in discussion The user's answers and decisions made while working. Omit until there are some. A decision that departs from the spec is recorded as `DEVIATION FROM SPEC (<date>): <what we skip or do otherwise>. Spec: «<quote>». User's answer: «<verbatim>»`; without a verbatim answer it stays in «Open questions». ## Open questions Contradictions and gaps. Omit when there are none. ## Links - Jira: <KEY> · Parent: <KEY> — <title> · pages, mockups and attachments, one line each ``` **Close to verbatim** means: keep the author's wording of every requirement — no paraphrasing, merging, softening or completing. Allowed: dropping greetings and noise, fixing markup, a bracketed note next to the original words («[typo? probably the admin role]»). A later comment replaces an earlier one only when it says so; otherwise both stay and the conflict goes to «Open questions». A question whose answer could mean not doing, or doing otherwise, what the spec requires is marked «(deviation from spec)» and carries no proposed answer: until the user decides, the spec applies. Afterwards only «Decided in discussion» and «Open questions» are ever edited. Tell the user the path. ## Rules - The deliverable is ALWAYS the file. Never treat writing it as optional. - READ-ONLY on Jira and Confluence: the client refuses anything but GET and JQL search. - Do not invent requirements. If it is undecided, ask; do not guess. - When comments conflict, the user's answers win.
Pliki w paczce (3)
- SKILL.md 5 KB
- scripts/fetch_issue.py 6 KB
- scripts/jira.py 10 KB
Wersje
W bazie nie ma innych skilli o nazwie jira-reader.
Migawka bazy z 13.11.2026 · wersja paczki 300191a7cf6d (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100confluence-jira
Integruje Confluence z JIRA: osadza zagadnienia na stronach i tworzy powiązania między obydwoma produktami Atlassian. - 90jira-project-management
Obsługuje zgłoszenia Jira i dokumenty projektowe w Confluence dla zespołu: statusy sprintów, komentarze i zmiany stanu. - 90confluence
Wyszukuje, czyta, szkicuje, tworzy i aktualizuje strony wiki Confluence Cloud: tytuły, sekcje, przestrzenie i strony podrzędne. - 100atlassian
Zarządza zgłoszeniami Jira i stronami Confluence w Atlassian Cloud: wyszukiwanie i tworzenie przez JQL i CQL, workflowy, przejścia, komentarze, z OAuth 2.1 przez serwer MCP lub tokenem API. - 100jira-assistant
Zarządza zgłoszeniami Jira przez Atlassian MCP: wyszukiwanie, tworzenie, aktualizacja, zmiana statusu i zadania sprintu, z automatycznym wykrywaniem konfiguracji obszaru roboczego. - 100atlassian-content
Tworzy dobrze sformatowane zgłoszenia i komentarze w Jira oraz strony w Confluence na Cloud i Server, z obowiązkowym ponownym sprawdzeniem i naprawą po utworzeniu.