Break Down to Sub-tasks — Autonomous Agents Agentic Workflow

schema_version: 1
id: agent_pm_breakdown
title: "Break Down to Sub-tasks"
description: "Decompose an epic into child issues with dependency waves (using the board's own issue types), and reconcile them (create, update, remove) to match the thread — issues only, no portal Tasks, no dispatch"
overview: |
  A planning pass that turns an analyzed epic into its child issues (the ticket is usually the epic
  itself). The PM agent reads the epic and its ENTIRE comment thread — the analysis, the decisions and
  answers, and any prior context — grounds itself in the relevant skill packs and the codebase, then
  creates independently-shippable Jira issues under the epic (as many as the work needs, using the
  board's own issue types), each with acceptance criteria, a workspace, a suggested delivery playbook,
  and a size, linked by blocked-by / blocks into parallel waves.

  It is idempotent: an epic with no children gets a fresh set; an epic that already has children is
  reconciled to the latest decisions and answers — issues are added, updated, or removed — never
  duplicated. It maintains Jira issues ONLY — moving an issue into the portal Tasks pool is a separate
  one-click action, and engineering work runs there, not here. The verdict reports whether issues were
  created fresh or reconciled. Use it as the breakdown step of the goal-shaping flow, or standalone on
  any analyzed epic.

category:
  level: pipeline
  domain: engineering

metadata:
  agent: true
  role: pm
  # Gate-verdict vocabulary for portal playbook steps (PLAYBOOKS.md §5) — matched from the
  # step's ticket comment; declaring it here is what makes this workflow content-routable.
  verdicts: [ISSUES_CREATED, ISSUES_RECONCILED]

params:
  agentic_app:
    type: string
    default: "cc"
    description: "Agent-app slug selecting the per-run env file ~/.agent-env.d/<slug> (AAP-C); 'cc'/subscription clears provider vars"
  ticket_url:
    type: string
    description: "Jira epic URL (the ticket to break down; usually the epic itself)"
  hint:
    type: string
    default: ""
    description: "Optional extra instructions for the agent"
  entry_path:
    type: string
    default: "~/workspace"
    description: "Working directory for the agent"

steps:
  - type: terminal.open
    title: "Agent: PM Break Down"
    cwd: "{{entry_path}}"
  - type: terminal.run
    cmd: |
      source ~/.agent-env
      # AAP-C (SCRUM-1506) + AAP-17 (SCRUM-1653): clear EVERY provider var an agent-app can deliver so
      # none hijacks/poisons a subscription run. A stray global ANTHROPIC_MODEL / ANTHROPIC_SMALL_FAST_MODEL
      # needs no CLAUDE_CODE_USE_* flag, so the old ${!CLAUDE_CODE_USE_@} glob never caught it — it survived
      # into the run and 404-ed aux/small-fast calls against api.anthropic.com. This explicit list mirrors
      # AGENT_APP_ENV_KEYS 1:1 (the-assistant website/src/lib/cloud/agent-app-env.ts — the single source of
      # truth; a parity test in each repo guards the two from drifting). Explicit over a glob: the union has
      # non-ANTHROPIC_ members (AWS_REGION, AWS_PROFILE, CLOUD_ML_REGION, CLAUDE_CODE_MAX_OUTPUT_TOKENS) and
      # a ${!AWS_@} glob would over-clear unrelated creds. Then source the per-run app env by slug when it
      # exists; no file => subscription/default. base/19-secrets stages ~/.agent-env.d/<slug>.
      unset \
        ANTHROPIC_API_KEY ANTHROPIC_BASE_URL ANTHROPIC_AUTH_TOKEN CCR_CONFIG_B64 \
        CLAUDE_CODE_USE_BEDROCK AWS_REGION AWS_PROFILE ANTHROPIC_MODEL \
        ANTHROPIC_SMALL_FAST_MODEL_AWS_REGION ANTHROPIC_SMALL_FAST_MODEL CLAUDE_CODE_MAX_OUTPUT_TOKENS \
        CLAUDE_CODE_USE_VERTEX CLOUD_ML_REGION ANTHROPIC_VERTEX_PROJECT_ID ANTHROPIC_VERTEX_BASE_URL \
        CLAUDE_CODE_USE_FOUNDRY ANTHROPIC_FOUNDRY_RESOURCE ANTHROPIC_FOUNDRY_BASE_URL \
        ANTHROPIC_DEFAULT_OPUS_MODEL ANTHROPIC_DEFAULT_SONNET_MODEL ANTHROPIC_DEFAULT_HAIKU_MODEL
      if [ -f "$HOME/.agent-env.d/{{agentic_app}}" ]; then
        source "$HOME/.agent-env.d/{{agentic_app}}"
      fi
      claude --dangerously-skip-permissions "$(cat <<'SB_PROMPT'
      read the ticket with attachments and ALL comments - {{ticket_url}}; it is usually the epic itself. if it fails, stop and report the error.
      this pass creates and maintains JIRA ISSUES ONLY - not the portal Tasks pool (a separate one-click action).
      learn the area first - ground yourself in the relevant skill packs and the codebase before decomposing.
      following the analysis, decisions, and answers in the thread, decompose the epic into as many independently-shippable child issues as the work needs - do not cap the count; size each so it can ship on its own. use the same issue types the board already uses. for each issue give:
        - TITLE: outcome-focused
        - ACCEPTANCE CRITERIA: testable, one line each
        - WORKSPACE: the workspace / area it belongs to
        - SUGGESTED PLAYBOOK: the playbook that would deliver it (e.g. bug-fix, feature-impl). a documentation-page child - one that creates or refreshes a page on a docs site - always takes docs-page, and takes the board's documentation issue type where it has one.
        - SIZE: S / M / L
      a documentation-page child ALSO carries its page contract in the description, as a yaml block the delivery workflows read: site (target site id), mode (audit / skill-refresh / drift-fix / page-refresh / new-page / post-deploy-verify), routes (the pages), sources (the pack/module content sources to write from), elements (element numbers, on drift-fix), decisions (open decision flags it depends on). without that block the write step cannot resolve the page and stops for a decision.
      set blocked-by / blocks links so the issues form parallel WAVES (wave 1 = no blockers).
      be idempotent and reconcile to the latest decisions and answers: create what is missing, update issues that changed, and delete or close issues that are no longer needed or that the thread asked to drop - never create duplicates.
      {{hint}}
      post ONE summary comment to the ticket: the issues created, updated, or removed (key + title) and the wave map.
      end with one line - "VERDICT:" followed by exactly one of ISSUES_CREATED (created fresh child issues) / ISSUES_RECONCILED (added to, updated, or pruned existing children) - and use these uppercase tokens nowhere else in the comment.
      UNATTENDED RUN — this is a dispatched job: nobody is watching this terminal and nobody will answer, so never call AskUserQuestion and never end your turn by putting a question to the operator. Apply the policy stated above and carry it out as written, including where carrying it out is destructive — the call to run this job was made when it was dispatched, and it is not yours to re-take. Where that policy leaves a genuine choice, make it yourself and keep going; only where it is genuinely SILENT, prefer the option that does not destroy work nobody can recreate. Say what you decided and what you skipped and why — but INSIDE the report this prompt already specifies: wherever a report shape, a token vocabulary or a last-line verdict is stated above, that format still binds exactly as written, and this clause adds no section, no token and nothing after the last line. Questions the deliverable is MEANT to contain are content, not an unanswered prompt. Asking does not pause the job either: an unanswered prompt is auto-answered from the option ordering once the unattended wait budget lapses (KAN-204), so asking only burns ~15 minutes of the run and hands a real decision to a default.
      SB_PROMPT
      )"