Skip to content

v2.2.0 — jira_issues sweep collector

Choose a tag to compare

@fatherlinux fatherlinux released this 29 Sep 19:43
· 98 commits to main since this release
736d5dc

Added

  • jira_issues collector. Runs up to six named JQL queries per sweep, with
    {since} substituted by the sweep window's start as a JQL datetime in the
    window's zone, so created >= "{since}" reads "since the last report". A
    query without the placeholder is sent unchanged, which is how a still-open
    list reaches past the window.

    Each issue becomes a fixed-shape record: key, browse link, issue type,
    status, components, created, age_days. Where an issue was filed by a web
    intake form, the description's flat Field: value pairs are parsed into a
    contact block and one capped detail line — first occurrence of a name
    wins, so trailing free text cannot overwrite a structured answer. A
    hand-filed ticket parses to nothing and still renders from its summary.

    Queries run in order and share one 100-record budget. A query that fails is
    recorded and the rest still run. A flagged result keeps only the fields the
    collector generated or that come from a controlled vocabulary, withholding
    summary, detail and contact — an intake form carries a stranger's prose.

    The sweep profile must grant the Jira backend jira_search; no write tool is
    reachable from a sweep.

Spec: .specify/specs/003-jira-collector/spec.md