jira

name	jira
description	Jira project management skill for creating, updating, commenting, and linking tickets via Atlassian MCP. Use when managing Jira issues.

Jira

Purpose

This skill provides comprehensive Jira project management capabilities through the Atlassian MCP integration. It enables agents to create, update, search, comment on, and link issues programmatically. Use this skill whenever a task involves managing work items in Jira — whether creating new tickets, updating existing ones, querying backlogs, or orchestrating cross-issue relationships.

Issue Types & Hierarchy

Hierarchy (top → bottom)

Epic
 └── Story / Task / Bug
      └── Sub-task

Issue Types

Type	Purpose	When to Use
Epic	Large body of work spanning multiple sprints	Feature areas, initiatives, major deliverables
Story	User-facing functionality	"As a user, I want..." — delivers value to end users
Task	Technical work not directly user-facing	Infrastructure, refactoring, tooling, research spikes
Sub-task	Granular unit of work under a Story/Task	Backend endpoint, frontend component, test suite, migration
Bug	Defect in existing functionality	Something that worked before and is now broken, or doesn't match spec

Hierarchy Rules

• Epics contain Stories, Tasks, and Bugs (via Epic Link field).
• Stories, Tasks, and Bugs can have Sub-tasks.
• Sub-tasks cannot have children.
• Every Story/Task/Bug should belong to an Epic unless it is truly standalone.

Fields Reference

Required Fields

Field	Description	Notes
project	Project key (e.g., PROJ)	Must exist in the Jira instance
summary	One-line title	Imperative mood, <80 characters
issuetype	Epic, Story, Task, Sub-task, Bug	Must match project's configured types

Common Optional Fields

Field	Description	Example Values
description	Detailed body (ADF or markdown)	See templates below
priority	Urgency level	Highest, High, Medium, Low, Lowest
labels	Categorical tags (array)	["backend", "api", "tech-debt"]
components	Architectural components (array)	["auth-service", "web-app"]
assignee	Account ID of the assignee	Use jira_search to resolve names → IDs
sprint	Sprint ID (not name)	Retrieve via board/sprint API
story_points / customfield_XXXXX	Estimation	Fibonacci: 1, 2, 3, 5, 8, 13
epic_link / customfield_XXXXX	Parent Epic key	PROJ-42
fix_versions	Target release versions	["v2.1.0"]

Summary Writing Conventions

• Imperative mood: "Add user authentication" not "Added user authentication" or "Adding user authentication"
• Under 80 characters: Be concise but specific
• No trailing period: Summaries are titles, not sentences
• Include scope hint: "[API] Add rate limiting to /users endpoint"
• Bug summaries state the symptom: "Login fails with 500 when email contains +"

Description Templates

Story Description:

h2. User Story
As a [persona], I want [action] so that [benefit].

h2. Acceptance Criteria
* [ ] Criterion 1
* [ ] Criterion 2
* [ ] Criterion 3

h2. Technical Notes
* Implementation approach or constraints
* Relevant endpoints, schemas, dependencies

h2. Out of Scope
* What this story explicitly does NOT cover

Bug Description:

h2. Summary
Brief description of the defect.

h2. Steps to Reproduce
# Step 1
# Step 2
# Step 3

h2. Expected Behavior
What should happen.

h2. Actual Behavior
What actually happens. Include error messages, status codes, screenshots.

h2. Environment
* Browser/OS/Device:
* Version/Build:
* Environment: staging / production

h2. Severity Assessment
* Impact: [Critical / High / Medium / Low]
* Frequency: [Always / Often / Sometimes / Rarely]
* Workaround: [None / Exists — describe]

Workflow States

Standard Workflow

To Do → In Progress → In Review → Done

State	Meaning	Transition Trigger
To Do	Work not started, in backlog or sprint	Issue created or moved to sprint
In Progress	Actively being worked on	Developer starts work
In Review	Code complete, awaiting review	PR opened or review requested
Done	Accepted and complete	PR merged, QA passed

Transition Rules

• Only transition forward unless explicitly reverting (e.g., review rejection → In Progress).
• Moving to "In Progress" should set/confirm the assignee.
• Moving to "Done" should verify all sub-tasks are also Done.

Issue Link Types

Link Type	Forward Description	Reverse Description	When to Use
Blocks	blocks	is blocked by	Issue A must complete before B can start
Relates to	relates to	relates to	Loosely related work, shared context
Duplicates	duplicates	is duplicated by	Same defect reported twice
Clones	clones	is cloned by	Copy of an issue for another project/sprint
Causes	causes	is caused by	Root cause relationship (bugs)

MCP Tools Available

jira_create_issue

Create a new Jira issue.

Parameters: project, summary, issuetype, description?, priority?, labels?, components?, assignee?, parent? (for sub-tasks)
Returns: key, id, self (URL)

jira_update_issue

Update fields on an existing issue.

Parameters: issue_key, fields (object with fields to update)
Returns: confirmation

jira_add_comment

Add a comment to an existing issue.

Parameters: issue_key, body (comment text)
Returns: comment ID, created timestamp

jira_link_issues

Create a link between two issues.

Parameters: inward_issue, outward_issue, link_type
Returns: confirmation

jira_search

Search issues using JQL.

Parameters: jql, fields? (array of field names), max_results?
Returns: array of matching issues

jira_get_issue

Retrieve full details of a single issue.

Parameters: issue_key
Returns: full issue object with all fields

jira_transition_issue

Move an issue to a new workflow state.

Parameters: issue_key, transition_id or transition_name
Returns: confirmation

JQL Quick Reference

Common Queries

-- My open issues
assignee = currentUser() AND status != Done

-- Sprint backlog
sprint in openSprints() AND project = PROJ

-- Unresolved bugs by priority
project = PROJ AND issuetype = Bug AND status != Done ORDER BY priority ASC

-- Recently updated
project = PROJ AND updated >= -7d ORDER BY updated DESC

-- Blocked work
issuefunction in hasLinks("is blocked by")

-- Epics missing stories
issuetype = Epic AND project = PROJ AND NOT issuefunction in hasLinks("is Epic of")

-- Issues without estimates
project = PROJ AND issuetype in (Story, Task) AND story_points is EMPTY AND sprint in openSprints()

JQL Operators

Operator	Example	Notes
=, !=	status = "In Progress"	Exact match
in, not in	issuetype in (Story, Bug)	Multiple values
~	summary ~ "auth"	Contains text (fuzzy)
is EMPTY / is not EMPTY	assignee is EMPTY	Null checks
>=, <=	created >= -30d	Date/number comparison
was, was in, was not	status was "In Progress"	Historical state
changed	status changed FROM "To Do" TO "In Progress"	Transition history
ORDER BY	ORDER BY priority ASC, created DESC	Sorting

JQL Functions

Function	Purpose
currentUser()	The authenticated user
openSprints()	All active sprints
closedSprints()	All completed sprints
futureSprints()	All upcoming sprints
startOfDay(), endOfDay()	Day boundaries
startOfWeek(), endOfWeek()	Week boundaries
now()	Current timestamp

Comment Conventions

• Status updates: Start with the new state. "In Progress — starting backend implementation."
• Questions: Tag the person. "@john.doe — can you clarify the auth flow for SSO users?"
• Blockers: Flag clearly. "BLOCKED — waiting on PROJ-456 (API schema finalization)."
• Completion: Summarize what was done. "Done — implemented rate limiter with token bucket algorithm. See PR #234."
• Technical notes: Use code blocks for snippets, stack traces, or config.

Conventions

1. Always search before creating — avoid duplicates. Run a JQL query for similar summaries first.
2. One ticket, one concern — don't bundle unrelated work into a single issue.
3. Link related issues — if two tickets touch the same code or feature, link them.
4. Keep summaries scannable — a PM should understand the ticket from the summary alone.
5. Set priority explicitly — never leave priority as the default unless it truly is Medium.
6. Assign during sprint planning — not at creation time, unless the assignee is obvious.
7. Sub-tasks for decomposition — if a Story has >3 days of work, break it into Sub-tasks.

Knowledge Strategy

• Patterns to capture: Project-specific field mappings (custom field IDs), common JQL queries that prove useful, description templates refined for specific teams.
• Examples to collect: Successfully created ticket payloads, effective bug reports, JQL queries for recurring audits.
• Update permission: Agents may freely add/update files in references/. Changes to SKILL.md or scripts/ require user approval.