Jira Administration

SkillProductivity

Run bounded Jira administration workflows for project Components or Atlassian instance migration. Use for jira-components, sync Jira components, jira-instance-migration, changed Jira URL, or repoint Jira. Both modes are sealed behind read-first analysis and explicit user approval before any Jira, credential-session, or repository mutation.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Jira Administration skill

What this skill tells your AI

The instructions your AI receives, as published by upex-galaxy/agentic-qa-boilerplate in .agents/skills/jira-administration/SKILL.md and read by ahel’s review.

Mode routing

Choose exactly one mode and load only its reference.

ModeLegacy alias / triggerReference
componentsjira-components, reconcile/sync Jira Componentsreferences/components.md
instance-migrationjira-instance-migration, changed/repoint Jira instancereferences/instance-migration.md

If the mode is unclear, ask. Never combine both modes in one run.

Compact Rules

The sealed mutation contract. Binding on every run of either mode:

  • Exactly ONE mode per run: components (references/components.md) or instance-migration (references/instance-migration.md). Load only that mode's reference. Never combine the two, never fall through into the other.
  • Mode unclear → ASK. Do not infer one from a bare "fix Jira" / "sync Jira" request.
  • Load /acli before any Jira operation. Load other tool-owner skills only when the selected reference requires them.
  • Missing MCP or Jira credentials = HARD STOP (AGENTS.md Critical Rule #10). Name the exact env var, point at .env / .env.example, ask for an agent-session restart. No workaround, no partial run.
  • Read-first on every mutation: inspect the live state before authoring any plan. Nothing is created, applied, deleted, or repointed without the user's explicit approval given inside the same run.
  • components: derive and inspect → author the plan file → dry-run → WAIT for explicit approval → only then --apply.
  • instance-migration: resolve and confirm BOTH instances → audit and verify reachability → WAIT for explicit approval → only then change files or the acli session. That session lives at ~/.config/acli and is machine-global: re-login repoints every repo on the host, not just this one.
  • The Atlassian host lives in .agents/project.yamlissue_tracker.atlassian_url and NOWHERE else locally. A stale ATLASSIAN_URL in .env or the process environment is contamination to DELETE, never to update — a second copy is what goes stale.
  • Template-repo carve-out: if .agents/project.yamlproject.project_name is null, the repo is an un-onboarded template. Leave atlassian_url and project_key null, say so in the report, and never manufacture a commit to hide the emptiness.
  • Run only the selected reference's verification steps. Never run the other mode's.
  • Forward $ARGUMENTS unchanged.

Read full SKILL.md when: the mode is ambiguous, a dry-run diff or migration audit looks wrong, or you need the selected reference's step-by-step phases and verification list.

Signals

GitHub stars
21
Forks
12
Last commit
Sep 2026

ahel review

  • S4info
    community integration — published by upex-galaxy, not jira

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Catalog kind
skill
Gateway key
jira-administration
Source
github.com/upex-galaxy/agentic-qa-boilerplate