| name | jira |
| description | Tools for interaction with jira production cloud environment. Use when asked to: search or filter issues with JQL, create or update tickets, transition issue status, manage epics and sprints, log work, add comments, or generate project reports. Trigger keywords: jira.
|
| compatibility | Requires jira agent (jira.agent.md) and jira MCP server. |
JIRA Environment Skill
Agent Delegation
Delegate ALL operations to the jira agent. The agent provisions the Jira MCP server
and has the required tools for all read and write operations.
Description
Access Jira Cloud to read and manage issues, epics, sprints, boards, projects, and worklogs.
Use for any task involving Jira tickets, sprint planning, status updates, linking issues, or reporting.
Capabilities
- Search issues with JQL (Jira Query Language)
- Get, create, update, and delete issues
- Transition issue status (e.g., To Do → In Progress → Done)
- Add comments and worklogs to issues
- Manage epics and link issues to epics
- Create and link issues (Blocks, Relates to, Duplicate, etc.)
- List and search projects
- Get and manage agile boards and sprints
- Batch create multiple issues at once
- Download issue attachments
- Get changelogs and history for issues
- Manage fix versions / releases
Activation
Include jira in your prompt to activate this skill.
Agent Constraints
When connected to the Jira MCP server the agent MUST enforce:
- Default project: use project key
PD when the user does not specify a project — never override an explicit project key
- cloudId: use
"https://${JIRA_ORG}.atlassian.net" — do NOT call getAccessibleAtlassianResources
- Search page size: default
maxResults: 10 / limit: 10; paginate with startAt when a task requires more results
Connectivity Check
Before taking any action the agent MUST:
- Call
jira_get_all_projects with maxResults=1 as a lightweight connectivity probe.
- If the probe fails, stop immediately and report: "Jira MCP server is unavailable. Cannot proceed."
- Only proceed after a successful probe response.
Usage Examples
Search Issues with JQL
jira Find all open bugs assigned to me in project PD
Sprint Overview
jira Show all issues in the current active sprint
Create an Issue
jira Create a bug in project PD: "Login fails on Safari" assigned to john@example.com
Transition Status
jira Move PD-123 to In Progress
Add Comment
jira Add comment to PD-456: "Fixed in PR #89, pending review"
Epic Management
jira Show all stories linked to epic PD-10
Log Work
jira Log 2h 30m on PD-789 with comment "Implemented feature"
JQL Reference
Common JQL patterns (replace PD with the target project key as needed):
assignee = currentUser() AND status != Done
project = PD AND issuetype = Bug AND priority = High
sprint in openSprints() AND status = "In Progress"
updated >= -7d AND project = PD
parent = PD-123 (child issues under an epic — field availability varies by instance; use jira_search_fields to verify)
Configuration
- Do NOT read filesystem config files (
mcp-config.json, ~/.claude/mcp-config.json, etc.)
- Always route through the
jira.agent.md custom agent which owns the MCP server config.
- Auth token is read from
JIRA_TOKEN env var defined in .claude/.env.
Limitations
- Jira Cloud only (not Server/Data Center for all endpoints)
- Write operations require appropriate Jira permissions
- Custom field IDs vary per instance — use
jira_search_fields to discover them before use