jira-issue-trackerlisted
Install: claude install-skill thientranhung/astragentic
# Tracker adapter: Jira
**Measured on a live project that moved Linear → Jira on 2026-08-21**, team-managed. Every trap below cost that project a run or a correction. What the pipeline
requires of *any* tracker is `.agents/tracker-contract.md`; this file is the HOW for one of them.
**The project half stays in the project.** This adapter carries no site, no `cloudId`, no project
key, no transition-id table — the project's `docs/agents/issue-tracker.md` owns those, and they
are per-project by nature. Substitute as you read.
Reach it through the **Atlassian MCP tools** in a session, and through **`twg`** from a shell.
## The shell route: `twg` (4.4.0)
The Teamwork Graph CLI reads Jira with the login it keeps per machine (`~/.config/twg`), so the
plugs that used to print `unreachable -` can answer. Declare `jira <PROJECT-KEY>` in
`.astraler/project/tracker` and `scripts/tracker.sh` does the reading for `ticket-done.sh`,
the session-start counts and the dispatch mod; delete the old `unreachable` plugs, because a
plug wins over the declaration. Measured 2026-10-24, on a live project:
| command | time | cost |
|---|---|---|
| `twg jira workitem get <KEY-n> -o json --output-summary none` | 1.5 s | native, no credit |
| `twg jira workitem query --jql '<jql>' --limit 500 -o json --output-summary none` | 5.4 s (22 tickets) | native, no credit |
| anything `twg --help` marks `[Enriched]`, and `twg rovo …` | — | Rovo credits: ask the owner first |
**The state is the status CATEGORY