Jira Data Center CLI with SKILLS.

jira-cli gives coding agents and developers a concise, deterministic interface to Jira Data Center. It hides REST details, normalizes noisy Jira responses, and keeps commands discoverable through --help.

The CLI works on its own and publishes an optional first-party Agent Skill for more reliable workflows across supported coding agents.

Calling Jira directly from a coding agent works, but often requires large REST responses, Jira-specific pagination and field knowledge, repeated authentication setup, and custom error handling.

jira-cli provides one reusable boundary between agents, scripts, and Jira:

coding agent / script
         |
         v
     jira-cli
         |
         v
Jira Data Center REST API

Commands return compact, predictable results so agents can spend their context on the task instead of Jira transport details.

• Provide stable support for Jira Data Center 9.12.x.
• Keep commands concise and discoverable through --help.
• Produce deterministic, schema-friendly output and errors.
• Reduce token usage through compact defaults, field selection, and bounded pagination.
• Keep application logic independent from the command-line framework.
• Ship a first-party Skill and reusable core for future integrations.

• Replacing the Jira web interface.
• Supporting Jira Cloud.
• Covering every Jira REST or administration endpoint.
• Generating JQL with a built-in language model.
• Hiding Jira permissions, workflow rules, or field configuration.

• Jira Data Center 9.12.x.
• Jira Core REST API 9.12.14.
• Node.js 20 or newer on Linux, macOS, Windows, and WSL.
• Personal Access Token authentication over HTTPS.

• Token-efficient normalized Jira output.
• Agent-ready command help and installable Skill.
• Safe mutations with exact confirmation for destructive operations.
• Multiple output formats for agents, humans, and shell automation.
• Predictable errors and exit codes for automation.
• Raw output escape hatch for instance-specific fields and integrations.

• CLI: Best suited to coding agents and terminal workflows. Commands expose a small, stable interface and return compact JSON without loading large API tool schemas or raw Jira payloads into the model context.
• REST API: Better for custom applications that need endpoints or response fields outside the CLI contract. Use --output raw before dropping down to direct REST calls.

• Node.js 20 or newer.
• Jira Data Center with Personal Access Token support.
• A coding agent such as Codex, Claude Code, or GitHub Copilot is optional.

npm install --global @tquoctuan97/jira-cli@latest
jira-cli --help

jira-cli auth login
jira-cli auth status

auth login accepts only HTTPS Jira URLs. The PAT is entered without terminal echo and stored in ~/.config/jira-cli/credentials.json on Linux and WSL. The file is plaintext and protected with filesystem permission 0600. During interactive login, the CLI shows a link to the Jira Personal Access Token page after the base URL is entered.

For automation, provide credentials without command-line arguments:

printf '%s' "$JIRA_TOKEN" |
  jira-cli auth login --base-url "$JIRA_BASE_URL" --token-stdin

The CLI and Agent Skill are installed separately: the CLI executes Jira operations, while the Skill teaches coding agents how to use it safely and efficiently.

List the skills available in this repository:

npx skills add tquoctuan97/jira-cli --list

Install jira-cli for coding agents detected in the current project:

npx skills add tquoctuan97/jira-cli --skill jira-cli

Install it globally for a specific agent:

npx skills add tquoctuan97/jira-cli \
  --skill jira-cli \
  --agent codex \
  --global

Use --yes for non-interactive installation. See the skills CLI for supported agents and installation options.

Point your agent at the CLI and let it discover commands through help:

Find my unresolved Jira issues and summarize the three most recently updated.
Use jira-cli and check jira-cli --help for available commands.

# Read a compact issue view
jira-cli issue get FE-123 --fields key,summary,status,assignee

# Search with bounded pagination
jira-cli issue search \
  --jql 'project = FE AND statusCategory != Done ORDER BY updated DESC' \
  --fields key,summary,status,assignee,updated \
  --limit 20

# Discover and execute an exact workflow transition
jira-cli issue transitions FE-123
jira-cli issue transition FE-123 --to 31

# Add a comment
jira-cli comment add FE-123 --body 'Ready for review'

# Download an attachment without overwriting an existing file
jira-cli attachment download 10042 --output-file ./report.pdf

JSON is the default machine-readable format. The same command can return a human-readable summary or the original Jira response when needed:

jira-cli issue get FE-123                         # normalized JSON
jira-cli issue get FE-123 --output markdown       # agent-friendly Markdown
jira-cli issue get FE-123 --output text           # concise terminal output
jira-cli issue get FE-123 --output raw            # original Jira response

Use --fields to return only the data needed for the task, or --output-file to keep large results out of the agent context:

jira-cli issue search \
  --jql 'project = FE ORDER BY updated DESC' \
  --fields key,summary,status \
  --output-file ./issues.json

jira-cli auth login
jira-cli auth status
jira-cli auth logout

jira-cli issue get <issue-key>
jira-cli issue search --jql <query>
jira-cli issue create [field-options]
jira-cli issue create --input <file-or-dash>
jira-cli issue update <issue-key> [field-options]
jira-cli issue update <issue-key> --input <file-or-dash>
jira-cli issue delete <issue-key> --confirm <issue-key>
jira-cli issue create-meta --project <key> --type <name-or-id>
jira-cli issue edit-meta <issue-key>
jira-cli issue history <issue-key>

Simple field options include --project, --type, --summary, --description, --assignee, --priority, --labels, and --parent. Project is create-only.

jira-cli issue transitions <issue-key>
jira-cli issue transition <issue-key> --to <exact-id-name-or-status>
jira-cli issue assign <issue-key> --to <username-or-me>
jira-cli issue unassign <issue-key>
jira-cli issue link-types
jira-cli issue link <source-key> <target-key> --type <exact-id-or-name>
jira-cli issue unlink <link-id> --confirm <link-id>
jira-cli issue watch <issue-key>
jira-cli issue unwatch <issue-key>
jira-cli issue watchers <issue-key> [--include-users]

jira-cli comment list <issue-key>
jira-cli comment get <issue-key> <comment-id>
jira-cli comment add <issue-key> --body <text>
jira-cli comment update <issue-key> <comment-id> --body <text>
jira-cli comment delete <issue-key> <comment-id> --confirm <comment-id>
jira-cli attachment list <issue-key>
jira-cli attachment add <issue-key> <file> [additional-files...]
jira-cli attachment download <attachment-id> --output-file <path>
jira-cli attachment delete <attachment-id> --confirm <attachment-id>

Use --body-file <file-or-dash> for long comment bodies.

jira-cli project list
jira-cli project get <project-key>
jira-cli field list
jira-cli user find --query <text>
jira-cli user find --assignable-to <issue-key>

    --config <path>        Override the session configuration path
-o, --output <format>      json, raw, markdown, or text
-f, --fields <fields>      Comma-separated fields to return
    --output-file <path>   Write command output to a file
    --timeout <ms>         Request timeout
    --quiet                Suppress non-result diagnostics
    --verbose              Include safe diagnostics

Run command-level help for the authoritative interface:

jira-cli --help
jira-cli issue --help
jira-cli issue search --help

pnpm install
pnpm typecheck
pnpm test
pnpm build

Detailed product scope and architecture live in docs/.

MIT