skill-slide-planning

name	skill-slide-planning
description	Interactive slide planning with narrative arc review and per-slide feedback. Invoke for /plan on present:slides tasks.
allowed-tools	Agent, Bash, Edit, Read, Write, AskUserQuestion
context	fork
agent	slide-planner-agent

Slide Planning Skill

Interactive 5-stage Q&A skill for slide planning. Gathers theme preference, narrative arc feedback, slide include/exclude decisions, and per-slide refinement before delegating to slide-planner-agent for slide-by-slide plan generation.

IMPORTANT: This skill implements the skill-internal postflight pattern. After the subagent returns, this skill handles all postflight operations (status update, artifact linking, git commit) before returning.

Context References

Reference (do not load eagerly):

• Path: .claude/context/formats/return-metadata-file.md - Metadata file schema
• Path: .claude/context/patterns/postflight-control.md - Marker file protocol
• Path: .claude/context/patterns/file-metadata-exchange.md - File I/O helpers
• Path: .claude/context/patterns/jq-escaping-workarounds.md - jq escaping patterns (Issue #1132)

Note: This skill runs interactive Q&A then delegates to slide-planner-agent. Context is loaded by the agent.

Trigger Conditions

This skill activates when:

• /plan on a present task with task_type: "slides" (or compound present:slides)
• Manifest routing: plan -> present:slides -> skill-slide-planning
• Present extension is available

---

Input Parameters

Required Parameters

• task_number - Task number (must exist in state.json with task_type containing "slides")
• session_id - Session ID from orchestrator

Optional Parameters

• plan_path - Path to existing plan (for resume, not typically used)
• resume_phase - Phase to resume from (not typically used)

---

Execution Flow

Stage 0: Input Validation

Validate required inputs:

• task_number - Must be provided and exist in state.json
• Verify task_type contains "slides" (supports "present:slides", "slides", or legacy language="present" + task_type="slides")

# Lookup task
task_data=$(jq -r --argjson num "$task_number" \
  '.active_projects[] | select(.project_number == $num)' \
  specs/state.json)

# Validate exists
if [ -z "$task_data" ]; then
  return error "Task $task_number not found"
fi

# Extract fields
task_type=$(echo "$task_data" | jq -r '.task_type // ""')
status=$(echo "$task_data" | jq -r '.status')
project_name=$(echo "$task_data" | jq -r '.project_name')
description=$(echo "$task_data" | jq -r '.description // ""')
forcing_data=$(echo "$task_data" | jq -r '.forcing_data // {}')

---

Stage 1: Preflight Status Update and Postflight Marker

Source skill-base.sh once, then follow @.claude/context/patterns/skill-preflight-flow.md in full for Stage 2 (preflight status update) and Stage 3 (marker creation):

source .claude/scripts/skill-base.sh
padded_num=$(printf "%03d" "$task_number")
task_dir="specs/${padded_num}_${project_name}"
mkdir -p "$task_dir"
skill_name="skill-slide-planning"
operation="plan"

---

Stage 2: Check for Existing Design Decisions

existing_dd=$(echo "$task_data" | jq -r '.design_decisions // empty')
if [ -n "$existing_dd" ]; then
  # AskUserQuestion:
  #   "Previous design decisions found for this task:
  #    Theme: {theme}
  #    Slides: {included count} included, {excluded count} excluded
  #    Feedback on {count} slides
  #
  #    Reuse existing decisions, or start fresh?"
  #
  # If "reuse": skip to Stage 7 (delegation)
  # If "start fresh": clear design_decisions and continue
fi

---

Stage 3: Theme Selection (Interactive Stage 1)

Read the research report to extract the recommended theme:

report_path=$(ls -1 "${task_dir}/reports/"*_slides-research.md 2>/dev/null | sort -V | tail -1)
# Extract recommended theme from report (if exists)

AskUserQuestion:

{If report exists: "The research report recommends: {recommended_theme}\n\n"}
Select a visual theme for your {talk_type} presentation:

A) Academic Clean -- Minimal, high-contrast, serif headings. Best for department seminars.
B) Clinical Teal -- Medical/clinical palette, clean data presentation. Best for clinical audiences.
C) Conference Bold -- Strong colors, large type, designed for projection. Best for conference talks.
D) Minimal Dark -- Dark background, high contrast, code-friendly. Best for technical audiences.
E) UCSF Institutional -- Navy/blue palette, Garamond headings. Best for UCSF presentations.

Or type a custom theme description.

Map response to theme slug:

• A -> academic-clean
• B -> clinical-teal
• C -> conference-bold
• D -> minimal-dark
• E -> ucsf-institutional
• Custom text -> store as-is

Store as design_decisions.theme.

---

Stage 4: Narrative Arc Outline (Interactive Stage 2)

Build the narrative outline from the research report slide map (or talk pattern if no report):

For each slide in the map:

{position}. [{type}] {one-line content summary} ({REQUIRED|optional})

AskUserQuestion:

Here is the narrative arc for your {talk_type} talk ({duration} min, {slide_count} slides):

1. [title] "{talk_title}" -- authors, affiliations, date
2. [motivation] Gap in knowledge: {summary} (REQUIRED)
3. [background] Literature context: {summary} (REQUIRED)
...
{N}. [acknowledgments] Funding, collaborators: {summary} (optional)

Estimated timing: ~{duration/slide_count} min/slide

Feedback on the narrative arc:
- Reorder? (e.g., "move 10 before 9")
- Add slides? (e.g., "add a slide about X after 5")
- Remove slides? (e.g., "remove 8")
- Change emphasis? (e.g., "expand results to 3 slides")
- Or type "looks good" to proceed as-is.

Process feedback:

• Parse reorder instructions, insertions, removals, emphasis changes
• Rebuild the slide list with changes applied
• New slides from user additions get type "custom"

Store as:

• design_decisions.narrative_arc - Final ordered slide list (array of objects with position, type, summary, required flag)
• design_decisions.arc_feedback - Raw user feedback text

---

Stage 5: Slide Picker (Interactive Stage 3)

Present the updated slide list with 2-3 line content previews per slide.

AskUserQuestion:

Review each slide and mark for inclusion. Type the numbers of slides to EXCLUDE
(all are included by default), or "all" to include everything.

 1. [title] "{talk_title}"
    Content: Authors, affiliations, conference/date

 2. [motivation] Gap in knowledge
    Content: {2-3 line preview of mapped content}

 ...

 {N}. [acknowledgments] Funding and collaborators
    Content: {2-3 line preview}

Exclude slides (comma-separated numbers), or "all" to keep everything:

Parse response:

• "all" or empty -> include everything
• Comma-separated numbers -> exclude those positions

Store as:

• design_decisions.included_slides - List of included slide positions
• design_decisions.excluded_slides - List of excluded slide positions

---

Stage 6: Per-Slide Detail and Feedback (Interactive Stage 4)

For each included slide, show detailed content mapping and collect optional feedback. Uses a single consolidated AskUserQuestion to minimize round-trips.

For long talks (20+ included slides): Group by section and show section summaries first. Accept "section looks good" for entire groups.

AskUserQuestion:

Here are the {N} included slides in detail. For any slide you want to adjust,
type its number and your feedback. Examples:

  2: emphasize the clinical urgency more
  5: include the CONSORT diagram
  6: split into two slides, one for each primary outcome
  9: add comparison to the Smith et al. 2024 findings

---
Slide 1: [title] "{talk_title}"
  Content: {full content mapping from research report}
  Speaker notes: {suggested talking points}
  Template: {template name}

---
Slide 2: [motivation] Gap in knowledge
  Content: {full content mapping}
  Speaker notes: {suggested talking points}
  Template: {template name}

...

Enter feedback (one per line, "done" when finished, or "looks good" for no changes):

Parse response:

• "looks good" or "done" with no feedback -> no changes
• Lines matching {number}: {feedback} -> store per-slide feedback

Store as design_decisions.slide_feedback (map of slide position string -> feedback text).

---

Stage 7: Delegate to slide-planner-agent

Assemble the complete delegation context:

{
  "session_id": "{session_id}",
  "delegation_depth": 1,
  "delegation_path": ["orchestrator", "plan", "skill-slide-planning", "slide-planner-agent"],
  "task_context": {
    "task_number": "{task_number}",
    "task_name": "{project_name}",
    "description": "{description}",
    "task_type": "present:slides"
  },
  "research_report_path": "{report_path}",
  "design_decisions": {
    "theme": "{selected theme}",
    "narrative_arc": [{position, type, summary, included, feedback}],
    "arc_feedback": "{raw arc feedback}",
    "included_slides": [1, 2, 3, ...],
    "excluded_slides": [7, 8, ...],
    "slide_feedback": {"2": "feedback text", ...}
  },
  "forcing_data": "{from state.json task metadata}",
  "metadata_file_path": "{task_dir}/.return-meta.json"
}

CRITICAL: Use the Agent tool to spawn slide-planner-agent. Do NOT use Skill(...).

Tool: Agent (NOT Skill, NOT Plan)
Parameters:
  - subagent_type: "slide-planner-agent"
  - prompt: [Include full delegation context above]
  - description: "Create slide plan for task {N}"

Also store design decisions in state.json before delegation:

# Store design_decisions in state.json task metadata
# Use jq to update the task entry with the complete design_decisions object

---

Stage 7b: Self-Execution Fallback

CRITICAL: If you performed the work above WITHOUT using the Agent tool (i.e., you read files, wrote artifacts, or updated metadata directly instead of spawning a subagent), you MUST write a .return-meta.json file now before proceeding to postflight. Use the schema from return-metadata-file.md with the appropriate status value for this operation.

If you DID use the Agent tool, skip this stage -- the subagent already wrote the metadata.

---

Postflight (ALWAYS EXECUTE)

The following stages MUST execute after work is complete, whether the work was done by a subagent or inline (Stage 7b). Do NOT skip these stages for any reason.

Stage 8: Read Metadata File

metadata_file="${task_dir}/.return-meta.json"

if [ -f "$metadata_file" ] && jq empty "$metadata_file" 2>/dev/null; then
    meta_status=$(jq -r '.status' "$metadata_file")
    artifact_path=$(jq -r '.artifacts[0].path // ""' "$metadata_file")
    artifact_type=$(jq -r '.artifacts[0].type // ""' "$metadata_file")
    artifact_summary=$(jq -r '.artifacts[0].summary // ""' "$metadata_file")
else
    echo "Error: Invalid or missing metadata file"
    meta_status="failed"
fi

---

Stage 9: Update Task Status (Postflight)

Meta Status	Final state.json	Final TODO.md
planned	planned	[PLANNED]
partial	planning	[PLANNING]
failed	(keep preflight)	(keep preflight marker)

---

Stage 10: Link Artifacts

Add artifact to state.json with summary. Use the two-step jq pattern to avoid Issue #1132.

Update TODO.md: Link artifact per @.claude/context/patterns/artifact-linking-todo.md with field_name=**Plan**, next_field=**Description**.

---

Stage 11: Git Commit

Apply the plan scope from .claude/context/standards/git-staging-scope.md — targeted staging, never a repo-wide add: