plan-to-tickets

Plan to Tickets

Read one or more plans/active/<slug>/plan.md files. Decompose into Jira tickets. Write each as JSON under plans/active/<slug>/tickets/ so the draft state survives across sessions. Organize them into a target sprint (or backlog). User confirms; push via existing ticket flow. After push, archive the JSON drafts to archive/tickets-pushed/<YYYY-MM>/ for history.

Atlassian-enabled only.

References

• lib/folder-conventions.md - canonical home for ticket drafts (plans/active/<slug>/tickets/) and post-push archive (archive/tickets-pushed/<YYYY-MM>/)
• lib/plan-format.md - how to read plans; directory shape includes tickets/ subfolder
• lib/atlassian-rest.md - endpoints
• lib/comms-check.md - runs first
• lib/atlassian-enabled.md
• lib/jira-tags.md - apply cowork-planned, bootstrap
• Sibling: runesmith-jira:ticket (push flow)
• lib/user-prompts.md - structured-input requirement for any user prompt

User input rules

This skill follows the marketplace-wide user-prompt standard in lib/user-prompts.md. Every user prompt MUST use the host client's structured input UI (single-pick, multi-pick, or text-input form). Never freeform plain-text yes/no questions. The only exception is the consent-trigger gate documented in lib/consent.md, which waits for user-initiated phrases like "make the ticket".

Pre-Flight Checks

0. Comms check (always first)

1. Atlassian enabled

If not: "Atlassian not enabled. Run /atlassian:enable first."

2. Plans available

plans/active/ exists and has at least one plan with status: open or status: building.

Workflow

1. Pick plans

List all plans in plans/active/ with status: open or status: building. Show:

Plans available
  [1] {slug}  status: open  ({status sort})
  [2] {slug}  status: building
  ...

User picks one or more. Multiple plans can map to one batch of tickets.

2. Decompose

For each selected plan, propose a ticket breakdown. Strategy:

• Plan summary → Epic (if status: building and >5 child tickets, or user opts in).
• Plan acceptance criteria → Stories (one ticket per criterion, when criteria are well-formed).
• Implementation tasks - derive from plan body (Decision section + any "Implementation" section if the user added one). One Task per discrete work item.
• Bugs - only if the plan explicitly identifies known bugs to fix.

Show proposed tickets in a table. For each: Type (Epic / Story / Task / Bug), Summary, Description excerpt, target Sprint (default: backlog; user can pick active sprint).

3. Sprint targeting

For each ticket batch:

• Default → Backlog (no sprint id).
• User can route the whole batch into the active sprint or a future sprint.
• Mixed routing supported (some to active sprint, some to backlog).

If targeting a sprint, look up sprint id via GET /rest/agile/1.0/board/{BOARD_ID}/sprint?state=active (or future).

4. Show full draft + persist as JSON

Render every ticket as the JSON the runesmith-jira:ticket skill expects:

{
  "fields": {
    "project":   { "key": "{JIRA_PROJECT_KEY}" },
    "issuetype": { "name": "Task" },
    "summary":   "...",
    "description": <ADF document derived from plan section>,
    "labels":    ["cowork-planned", "bootstrap"],
    "priority":  { "name": "Medium" }
  }
}

Description ADF includes a link to the plan slug (back-reference).

Write each ticket JSON to plans/active/<slug>/tickets/<DRAFT-ID>.json where <DRAFT-ID> is a deterministic local identifier (e.g. <slug>-001, <slug>-002). These drafts persist on disk so the plan-to-tickets workflow can resume across sessions and so the user can review them outside chat.

If targeting a sprint, also include sprint assignment via POST /rest/agile/1.0/sprint/{SPRINT_ID}/issue after creates.

5. Get consent

{N} tickets to create:
  Epic:   1   {summary}
  Story:  3   ...
  Task:   5   ...
  Bug:    0

Target: {sprint name | Backlog}

Push these tickets to {JIRA_PROJECT_KEY}?

Wait for trigger phrase ("push", "make the tickets", "create them", "yes").

6. Push (delegated to ticket skill flow)

For each ticket, call the same REST flow as atlassian:ticket:

POST {SITE_URL}/rest/api/3/issue

Capture each response key.

After all tickets created, if any are sprint-targeted:

POST {SITE_URL}/rest/agile/1.0/sprint/{SPRINT_ID}/issue
{ "issues": [<list of new keys>] }

If Epic created, link Stories/Tasks to it via the Epic Link custom field (auto-discover via GET /rest/api/3/field).

7. Update plan

Append the new ticket keys to the plan's frontmatter tickets: array. Edit plan.md in place (preserving the rest):

tickets:
  - {KEY-1}
  - {KEY-2}

Touch updated: to today's date. If plan was status: open, optionally bump to status: building (ask user via structured prompt).

8. Archive pushed drafts

Per lib/folder-conventions.md, after successful Jira push:

• Rename each pushed JSON from plans/active/<slug>/tickets/<DRAFT-ID>.json → plans/active/<slug>/tickets/<JIRA-KEY>.json (so the on-disk name matches the issued key).
• Move the renamed file to archive/tickets-pushed/<YYYY-MM>/<JIRA-KEY>.json for history.
• plans/active/<slug>/tickets/ ends empty (or holds only unpushed drafts if user partially confirmed).

If any push failed, leave its draft JSON in plans/active/<slug>/tickets/ for retry; record the failure in the report.

9. Report

✓ Tickets created from plan(s)
Plans:    {slug-1}, {slug-2}
Tickets:  {N} created
  {KEY-1}  Task   {summary}  → {sprint name | Backlog}
  ...

Sprint assignments: {n}
Plan(s) updated with ticket references.
Drafts archived: archive/tickets-pushed/<YYYY-MM>/

Next: /runesmith-sprint:start-sprint to hand the active sprint to CC, or
      /runesmith-sprint:sprint-status to see board state.

Guard Rails

• Comms check ran first
• Atlassian enabled
• At least one plan selected
• Ticket types match plan content (no bugs unless plan identifies them)
• All tickets carry cowork-planned + bootstrap labels
• Description ADF includes plan slug back-reference
• Draft JSON written to plans/active/<slug>/tickets/ BEFORE push consent
• User explicitly consented before push
• Each ticket POST succeeds before sprint assignment
• Plan(s) updated with new keys
• Pushed drafts archived to archive/tickets-pushed/<YYYY-MM>/<KEY>.json
• Failed pushes leave draft JSON in plans/active/<slug>/tickets/ for retry
• No duplicate ticket creation if plan was previously processed (skill checks plan's existing tickets: and asks before adding more)

Error Cases

No plans with status open or building: "No active plans to convert. Create one with /core:plan first." Plan body missing acceptance criteria: Skill asks user to add them inline before decomposition, or proceeds with Task-only breakdown. Sprint id lookup fails: Default to Backlog; surface the failure but continue. Ticket create 400 (custom field required): Surface field name; ask user to fill or use a Story type that doesn't require it. Plan already has tickets[] populated: Show existing keys; ask "Add more, replace, or skip?" Sprint full / capacity warning: Note in report, doesn't block.