story-coverage

Story Coverage

Derive expected user stories from personas + goals + feature domain. Cross-reference NFR/GTM/rollout sections for orphan references.

When to Use

• shield:agile-coach calls this skill while grading dim 4 eval points 4f and 4g of a PRD
• /prd command (Phase B) calls this skill between PRD Sections 6 (Goals) and 8 (Stories) to scaffold expected stories

Input contract

The caller provides:

• personas: list of {id, name, description, goals}
• goals: list of {id, description}
• feature_domain: best-effort domain hint (e.g., "auth", "payment", "content", "internal-tool", "infrastructure")
• existing_sections: optional — content of NFR / GTM / Rollout / Risks sections (for orphan-reference detection)

Output contract

Return a list of expected stories with structured metadata:

{
  "expected_stories": [
    {
      "rationale": "persona-goal" | "archetype" | "orphan-reference",
      "persona_id": "P1",
      "goal_id": "G1",
      "archetype": "password-reset",
      "story_title": "Anika resets her password",
      "story_template": { ... },
      "severity": "P0" | "P1" | "P2"
    }
  ]
}

Derivation rules

Rule 1 — Persona × Goal cross-product

For each (persona, goal) pair, derive at minimum:

• Happy path story — the persona achieves the goal successfully
• One named error/recovery path — what happens when the happy path fails (timeout, partial failure, user abandons)

If 0 stories address a (persona, goal) pair → flag P0 ("persona P's goal G has no story"). If exactly 1 story (happy path only) → flag P1 ("persona P's goal G has happy path but no error/recovery").

Rule 2 — Archetype match

Look up the feature domain in archetypes.md. If the domain matches one of: auth, payment, content, lifecycle, multi-region, billing, observability, retrieve its archetypal flow list.

For each archetypal flow that's NOT covered by an existing story → flag P1 or P2 based on the archetype's marked severity in archetypes.md.