make-jira-from-spec

make-jira-from-spec

Turn spec changes into Jira work items. Reads the .ai/spec/ changes from the current session or a PR, brainstorms the decomposition, creates or updates Epics and Stories, then estimates and risk-assesses every item.

Defaults

Setting	Value
Project key	OLS
Cloud ID	redhat.atlassian.net
Content format	markdown
SP field	customfield_10028
Risk Score field	customfield_10976
Epic Size field	customfield_10795
Max story points	5 (split if above)

Invocation

/make-jira-from-spec
/make-jira-from-spec https://github.com/org/repo/pull/123
/make-jira-from-spec https://github.com/org/repo1/pull/10 https://github.com/org/repo2/pull/20
/make-jira-from-spec OLS-1234

Arguments (all optional, can be combined):

• PR URL(s) — one or more PR URLs to fetch spec diffs from (spec changes often span multiple repos)
• Jira key — existing Epic or Story to update

Step 1: Gather Spec Changes

Resolve the spec changes using this priority:

1a. PR URL(s) provided

Fetch the diff from each PR with gh pr diff <URL>. Filter to files under .ai/spec/. When multiple PR URLs are given, collect spec diffs from all of them — spec changes often span multiple repos. Read the full content of each changed spec file.

Overlap detection. If the same spec file appears in multiple PR diffs, fetch the head revision from each PR (gh pr view <URL> --json headRefOid) and compare the file at each head. Flag the overlap to the user before proceeding — do not silently pick one version. Ask which PR's version to use as the baseline for decomposition.

1b. Session context (no PR URLs)

Find spec files changed in the current session across all repos in the workspace:

# From the workspace root
for repo in */; do
  git -C "$repo" diff HEAD -- .ai/spec/ 2>/dev/null
done

Also check for untracked new spec files:

for repo in */; do
  git -C "$repo" diff --cached -- .ai/spec/ 2>/dev/null
  git -C "$repo" ls-files --others --exclude-standard .ai/spec/ 2>/dev/null
done