/detailing-mode

Detailing Mode Reference

Flesh out a previously-roadmapped cycle: take stories that have only titles and add sparks, content direction, and visual direction per story. Fires when the user re-enters cycle-design with an existing planning cycle directory.

The orchestrator command (`commands/craft-cycle-design.md`) defines the routing - this reference contains the full mode flow.

---

Detailing Mode (Flesh Out Planning Cycles)

Entered when `/craft` routes here with an existing cycle directory name as arg. This mode adds sparks and detail to title-only stories created in Roadmap Mode.

Pre-check

If args are provided: 1. Check if `.craft/cycles/[arg]/` exists 2. Check if `cycle.yaml` has `status: planning` 3. If valid: set PLANNING_CYCLE, **capture cycle.yaml's `source_concept`** into context, and proceed to Step D1 4. If invalid: show error and offer cycle picker

Use **Read** to read `.craft/cycles/[arg]/cycle.yaml`. Parse `status:` AND `source_concept:` values.

If the file exists and `status == "planning"`:

${CLAUDE_PLUGIN_ROOT}/hooks/scripts/update-global-state.sh PLANNING_CYCLE "[arg]"

Remember the cycle's `source_concept` for the rest of this session - it determines routing for any NEW stories added during detailing (see Step D2b).

If the file doesn't exist or status is not `planning` → show error and offer cycle picker.

**Tracker issue check**: read `.craft/cycles/[arg]/cycle.yaml`. If it has a non-empty `source_issue`, Read `${CLAUDE_PLUGIN_ROOT}/commands/references/tracker-follow.md` with that cycle's directory and continue only through its `## Continue`. A cycle without a `source_issue` skips this and nothing below changes.

Step D1: Show Current State

Read all stories in the cycle. Check which have sparks, type tags, and creative tool output:

# Check story state
Glob ".craft/cycles/[cycle]/stories/*.md" → story_files

stories = []
for each story_file in story_files:
  Read story_file (limit: 50) → parse frontmatter + section headings
  - title: from frontmatter
  - has_spark: "## Spark" section has non-empty, non-comment content
  - type: from frontmatter `type` field (may be missing)
  - has_content_direction: "## Content Direction" section has non-comment content
  - has_visual_direction: "## Visual Direction" section has non-comment content
  stories.append(title, has_spark, type, has_content_direction, has_visual_direction)

**Batch type inference for stories with sparks but no type:**

If any stories have sparks but missing `type` field, infer the type from the spark content and present for confirmation:

> "I've inferred types for stories with existing sparks: > > | # | Story | Type (inferred) | > |---|-------|-----------------| > | 1 | [Title] | `ui` | > | 2 | [Title] | `technical` | > > Correct? (Say which to change, or confirm to update the files.)"

On confirmation, update each story's frontmatter to add the `type` field.

**Present the full status:**

> "**Cycle: [Name]** — Detailing Mode > > | # | Story | Spark | Type | Content | Visual | > |---|-------|-------|------|---------|--------| > | 1 | [Title] | ✓ | `ui` | — | — | > | 2 | [Title] | — | — | — | — | > | 3 | [Title] | ✓ | `technical` | ✓ | n/a | > > [N] stories need sparks. [M] stories are missing creative tool output."

The "Visual" column shows `n/a` for non-ui stories, `—` for ui stories missing Visual Direction, and `✓` for ui stories with it. The "Content" column shows `—` for any story missing Content Direction and `✓` for those with it.

Use **AskUserQuestion**:

question: "How do you want to detail these stories?"
header: "Approach"
options:
  - label: "Go through all stories"
    description: "Capture sparks for each story in order"
  - label: "Pick specific stories"
    description: "Choose which stories to detail now"
  - label: "Add more stories first"
    description: "Add new story titles before detailing"
  - label: "Rearrange/rename"
    description: "Reorder or rename existing stories"

**If user provides custom text:** Ask a clarifying AskUserQuestion to understand their intent.

Step D2: Capture Sparks

**Planning-source routing applies when cycle.yaml has `source_concept` set (captured in Pre-check).** For each story that needs a spark, apply the action-moment framing:

• **Planning-extraction moment** (the spark for this story should be drawn from the cycle's planning concept): Read `${CLAUDE_PLUGIN_ROOT}/commands/references/story-from-planning.md` and execute its phases against the cycle's source_concept. The protocol writes the spark content from the planning doc and adds `source_concept` + `source_concept_last_updated` to the story's frontmatter.

**CRITICAL: include the invocation marker** when executing the protocol. State explicitly in your context before running Phase 1:

> `INVOCATION: cycle-design - planning source already confirmed in cycle.yaml`

This marker is required for Phase 1's auto-resolve to fire. Without it, Phase 1 falls back to corpus-scan and asks the user to pick a concept.

• **Add-a-separate-story moment / fleshing out a non-planning-sourced story**: Use the existing Step 3 flow below.

For a freeform story (or when cycle.yaml has no source_concept), reuse the existing **Step 3** flow (Capture Each Story):

• **3a. Choose Your Path** — "Let's get creative" or "I know what I want" (with recommendation for `type: ui` stories)
• **3b. Spark** — Capture the essence
• **3c. Rough Acceptance** — Optional
• **3d. Dependencies** — What depends on what
• **3e. Save Story** — Update the existing story file with spark content (including `type` in frontmatter)

Stories that already have sparks are skipped unless user explicitly asks to revisit.

**Existing stories without source_concept are NOT retroactively stamped during detailing.** If an existing story is being detailed and the user signals that it's actually planning-sourced ("this one is from the company-onboarding concept"), the orchestrator MAY ask whether to p