Analyze Story

Parse a user story .md file and return structured scenario data ready for TC file creation.

Expected input format

User story files follow this structure:

# [Feature] Add user story and scenarios for <Feature Name>

**As a** <role>
**I want** <capability>
**So that** <benefit>

---

## Scenario A: <Scenario Name>

**Preconditions:**
- <precondition>

**Steps:**
1. <step>
2. <step>

**Expected Result:**
- <expected result>

**Test Data:**
- <data>

---

## Acceptance Criteria
- <criteria>

## Notes
- <note>

Parsing rules

feature_name

Derived from the H1 title line using this fallback chain (stop at first match):

1. If title matches # [Feature] Add user story and scenarios for <X> → extract <X>
2. If title matches # [Feature] <X> → extract <X>
3. Otherwise → use the full title text after any leading # and strip common words ("feature", "the", "a")

Convert the extracted name to kebab-case for use as the feature directory name.

• Example: "Service Type Management" → service-type
• Example: # Login Feature → login
• Example: # User Authentication → user-authentication

user_story

Extract the three fields:

• as_a — text after **As a**
• i_want — text after **I want**
• so_that — text after **So that**

scenarios

Each ## Scenario N: <name> heading starts a new scenario block.

For each scenario, extract:

• scenario_letter — the letter (A, B, C…) from the heading
• scenario_name — the full name after the colon
• preconditions — bullet list under Preconditions:
• steps — numbered list under Steps: (raw text, no selector yet)
• expected_result — bullet list under Expected Result:
• test_data — bullet list under Test Data: (empty list if section absent)
• tags — inferred automatically (see Tag inference below)

Tag inference

Infer @smoke or @regression from the scenario name and expected result. Never require explicit tags in the story file.

Assign @smoke when the scenario describes:

• The primary success path of a core feature action (create, delete, submit, save)
• The list/index page loading correctly (columns, buttons visible)
• A critical read operation (view details, search returns results)
• Keywords: "successfully", "happy path", "valid", "loads", "appears", "visible"

Assign @regression when the scenario describes:

• Input validation (empty, whitespace, too long, invalid format)
• Duplicate / uniqueness enforcement
• Cancel or dismiss flows (modal stays open, no change persists)
• Permission / access control (non-admin, unauthorised)
• Error handling or negative outcomes
• Secondary success paths (edit, rename, reorder)
• Pagination, sorting, filtering edge cases
• Keywords: "fails", "rejected", "blocked", "cannot", "disabled", "cancel", "error", "invalid", "duplicate", "permission", "non-admin", "whitespace", "empty", "unchanged", "preserved"

When ambiguous, lean toward @regression — @smoke should be reserved for the handful of tests that confirm the feature is alive.

TC number assignment

TC numbers are assigned sequentially per feature, starting from 001.

Before assigning, scan plans/{feature_name}/ for existing {feature}-TC-NNN.md files to find the highest existing number. New scenarios are numbered from max + 1.

If no TC files exist yet, start from {feature}-TC-001.

Output structure

{
  "feature_name": "service-type",
  "user_story": {
    "as_a": "Admin user",
    "i_want": "to manage service types",
    "so_that": "I can configure the system"
  },
  "scenarios": [
    {
      "tc_number": "service-type-TC-001",
      "scenario_letter": "A",
      "scenario_name": "Create Service Type (Happy Path)",
      "preconditions": ["User is logged in as Admin"],
      "steps": [
        "Navigate to /service-types",
        "Click \"New Service Type\" button",
        "Enter name in name field",
        "Click \"Save\" button"
      ],
      "expected_result": ["A success notification appears", "The modal closes"],
      "test_data": [{"Name": "Wire Transfer"}],
      "tags": ["@smoke"]
    }
  ],
  "acceptance_criteria": ["..."],
  "notes": ["..."]
}

Rules

• Never assign a TC number already present in plans/{feature}/ — scan for {feature}-TC-NNN.md files.
• If a scenario heading is missing the letter (e.g. ## Scenario: Name), assign letters alphabetically.
• If the file has no ## Scenario headings, return an empty scenarios array and log a warning.
• Preserve step text exactly as written — do not paraphrase or normalise.