netresearch/jira-skill / skills/jira-communication
jira-communication
FlaggedUse when handling Jira issues, sprints, boards, links, fields, worklogs, attachments, or users, or on any Jira intent without a key ("create/find a ticket", "pick a project"). Auto-triggers on Jira URLs and issue keys (PROJ-123). Also use when MCP Atlassian tools fail or are unavailable for Jira Server/DC.
Jira Communication
CLI scripts via uv run, all supporting --help, --json, --quiet, --debug.
Auto-Trigger
On Jira URL or issue key (PROJ-123), pick by intent — each is one call:
| Intent | Tool |
|---|---|
| triage / work on ticket | jira-issue.py work KEY |
| start QA review | jira-issue.py qa KEY |
| QA-fail follow-up | jira-issue.py qa-fail KEY |
| field-only lookup | jira-issue.py get KEY --fields ... |
| change status | jira-issue.py act KEY → jira-transition.py do |
| audit / sibling discovery / "am I reviewing my own work?" | jira-qa-gather.py KEY |
Auth issues → jira-setup.py. Anti-pattern: get + comment list — use the matching verb.
work flags a ticket you have not taken: when it is not assigned to you, or its status says nobody works on it (To Do category, a review status such as QA, or resolved), a !! CLAIM KEY notice on stderr (claim in --json) lists the commands — jira-issue.py update KEY --assignee me, then the start or reopen transition from jira-transition.py list KEY. Run them before your first comment, measurement or edit. jira-comment.py add warns once before posting when the ticket is neither yours nor in a working status; it still posts.
Scripts
Under ${CLAUDE_SKILL_DIR}/scripts/{core,workflow,utility}/.
Core: jira-issue.py, jira-search.py, jira-worklog.py, jira-attachment.py, jira-setup.py, jira-validate.py
Workflow: jira-create.py, jira-transition.py, jira-comment.py, jira-move.py, jira-sprint.py, jira-board.py, jira-version.py, tempo-account.py
Utility: jira-user.py, jira-fields.py, jira-link.py, jira-weblink.py, jira-worklog-query.py, jira-watchers.py, jira-qa-gather.py
Execution Style
Run directly. Scripts report ✓/✗. Destructive ops: --dry-run. Global flags before subcommand: jira-issue.py --json get PROJ-123.
Every write ends with the ticket's state
After a successful write, these commands print the ticket's state to stderr: jira-comment.py add/edit, jira-transition.py do/path, jira-issue.py update, jira-worklog.py add, jira-weblink.py add, jira-link.py create (FROM_KEY, which is --target in the named form), jira-attachment.py add and jira-create.py issue:
── PROJ-123 · Task · Open · resolution: – · assignee: unassigned
time: logged 15m / estimate – / remaining – · labels: –
(!) ticket is "Open" while you work on it: claim it (assign + start work)
Read the (!) lines and act on them before reporting the work done. They appear when the ticket is Open or Backlog while you write to it, unassigned while In Progress, assigned to someone else while you write to it, already done while you write to it, done without a resolution, carries a resolution while not done, or has more time logged than estimated. An unassigned ticket in a review status such as QA gets no hint: it waits for the team pool by design. After a transition or a create only the resolution and time checks run, because the status and assignee are what the command just set.
With --json the same facts arrive as a ticket_state field in the output object, beside the keys it already had: key, status, status_category, issuetype, resolution, assignee, assignee_is_you, time_spent_seconds, original_estimate_seconds, remaining_estimate_seconds, labels and hints, the last being the (!) lines as strings. Read ticket_state.hints as you would the (!) lines. jira-attachment.py add --json prints a list, so its state goes to stderr as text instead. --quiet prints no state.
The state costs one extra read per write (jira-comment.py add reads once more before posting, for its claim warning) and never fails the command: a lookup error prints one line, or becomes "ticket_state": {"key": …, "error": "<type>"} in JSON. JIRA_TICKET_STATE=0 turns it off everywhere, together with the warning jira-comment.py add prints before posting; the work claim notice stays. logged is Jira's own time tracking: time booked elsewhere shows only once it is synced into a Jira worklog.
Posting wiki markup rewrites and checks it first
Every --comment and --description option that writes wiki markup runs three gates before the write, all on by default, because text that renders wrong is silent — the API returns 2xx either way. That is all seven: jira-comment.py add/edit, jira-transition.py do --comment, jira-transition.py path --comment, jira-worklog.py add --comment, and the --description of jira-create.py issue and jira-issue.py update. A body smuggled in through --fields-json is not gated — that option writes raw fields by design. jira-version.py writes two --description fields that are NOT gated (create and update); whether Jira renders a version description as wiki markup at all is unverified, and its help string claiming it does may simply be wrong.
- Dashes that Jira would render as strikethrough are escaped.
\-prints as a plain hyphen, so the posted text reads as written; stderr names how many lines changed and shows the first five. (The one shape where the escape is visible is two macros written against each other with no space — the dash can land inside a link target. Ordinary prose does not reach it.)--no-auto-escapekeeps the markup verbatim — but on its own it does not post a deliberate-strikethrough-: the lint and the render check each still refuse the span. Use--no-auto-escape --forcefor that. - The markup and the ticket language are linted. Block tags used inline (
{code},{noformat},{quote},{panel}are block-level), unbalanced tag counts, German prose on an English-only project, and a reference to another comment by position ("the comment above", "Kommentar oben" — comment order is a per-user setting; write "the previous comment" or use a permalink) each abort the write.--forceturns the findings into warnings and posts anyway. - The text is rendered by the instance and refused if it comes back struck through. This costs one API call per post and catches what no local check can — an autolinked issue key creates a boundary that exists only on an instance where that key resolves. A resolved issue's key, which Jira draws struck through as status styling, is not reported.
--no-preflightskips it; an unreachable renderer warns once and posts anyway.
--force posts despite any of the three. The flags are spelled the same on each command. All seven also take --dry-run: the escape and the lint still run — the preview shows the text a real write would post — while the render call and the write do not. See references/comments.md for the details.
Basic Usage
uv run ${CLAUDE_SKILL_DIR}/scripts/core/jira-issue.py get PROJ-123
uv run ${CLAUDE_SKILL_DIR}/scripts/core/jira-search.py query "assignee = currentUser() AND status != Closed" -n 5 -f key,summary,status
uv run ${CLAUDE_SKILL_DIR}/scripts/core/jira-issue.py update PROJ-123 --assignee me --priority Critical
uv run ${CLAUDE_SKILL_DIR}/scripts/workflow/jira-comment.py add PROJ-123 "Comment text"
uv run ${CLAUDE_SKILL_DIR}/scripts/workflow/jira-comment.py add PROJ-123 "Comment text" --no-auto-escape --force # deliberate -strikethrough-
uv run ${CLAUDE_SKILL_DIR}/scripts/workflow/jira-transition.py do PROJ-123 "In Progress"
uv run ${CLAUDE_SKILL_DIR}/scripts/core/jira-worklog.py add PROJ-123 2h --comment "Work done"
uv run ${CLAUDE_SKILL_DIR}/scripts/workflow/jira-create.py issue PROJ "Summary" --type Task
Transitions:
listshows each transition's id and what its screen requires; pass the id todo— a name or a target status is not always unique, and an ambiguous one is refused rather than guessed. Terminal transitions: pass--resolution <value>(Done,Won't do); if rejected ("cannot be set"), retry without it —references/intent-verbs.md. Versions: readreferences/versions.mdbeforejira-version.py. Mentions: posting commands verify[~username](miss → suggestions);get/workprint usernames (references/fields-and-users.md).
Related Skills
jira-syntax: descriptions/comments use Jira wiki markup, not Markdown.
No editorializing
State what happened, not how good it is — references/no-editorializing.md.
References
references/jql-quick-reference.md,references/jql-cookbook.mdreferences/multi-profile.md—--profilereferences/troubleshooting.md— auth, 401/403references/issue-editing.md— edit, delete, clear fields,--fields-jsonreferences/creation.md— create,--parent, fields, admin-scope (project,tempo-account.py)references/comments.md— edit, delete, lint, body via-, restricted comments (--visibility)references/worklog.md—--started, ranges,--tempo-account,deletereferences/attachments.md— upload, downloadreferences/links.md— linksreferences/agile.md— sprints/boardsreferences/no-editorializing.md— no self-praisereferences/fields-and-users.md— custom field IDs, users, issue types, screen fields (jira-fields.py meta)references/watchers.md— watch, subscribe, list watchersreferences/versions.md— fix/affects versions, releases, version CRUDreferences/qa-gather.md— audit bundle (siblings, prose URLs, self-review check from changelog + worklog — not the assignee)references/intent-verbs.md—work / qa / qa-fail / act, exact transition names
Authentication
Cloud: JIRA_URL + JIRA_USERNAME + JIRA_API_TOKEN. Server/DC: JIRA_URL + JIRA_PERSONAL_TOKEN. Config via ~/.env.jira or ~/.jira/profiles.json.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |