@ztffn/presentation-generator-plugin 1.5.0 → 1.6.1

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "presentation-generator",
   "description": "Generate complete graph-based presentations from natural language briefs and project documents. Pipeline: content extraction, narrative design, deterministic graph generation, and visual styling.",
-  "version": "1.5.0",
+  "version": "1.6.1",
   "author": {
     "name": "Huma"
   },

package/agents/presentation-narrative.md

@@ -32,7 +32,7 @@ Read `_temp/presentation-content-brief.json`. Note any weak or missing fields.
 ### Step 3 — Select framework, design structure, write slide content
 
 Based on `goal` and `audience`, select the appropriate narrative framework from your frameworks skill. Then:
-1. Design the spine (main horizontal flow) of 4-7 nodes
+1. Design the spine (main horizontal flow) — use the duration-based spine length table in your frameworks skill to determine node count
 2. Design drill-down branches where content warrants optional depth
 3. Write detailed content for each slide following the quality guidance in your slide-content skill
 
@@ -55,6 +55,10 @@ Write a markdown file to `_temp/presentation-outline.md`. The format is a hard c
 - CONTENT headers for spine: `### SPINE N: Title`
 - CONTENT headers for drill-downs: `### SPINE N.M: Title`
 
+**Two formats that look reasonable but silently break the parser:**
+- `- **Io Thermal Storage: The foundation** — Phase-change iron...` ✗ — title must not be inside the bold markers; the N.M index must be
+- `### Io Thermal Storage: The foundation` / `**(Drill-down under Spine Node 2)**` ✗ — no freeform title headers; must be `### SPINE 2.1: Io Thermal Storage`
+
 ```markdown
 # Presentation Outline: [Deck Title]
 

package/hooks/hooks.json

@@ -43,6 +43,16 @@
             "timeout": 30
           }
         ]
+      },
+      {
+        "matcher": "Skill",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "${CLAUDE_PLUGIN_ROOT}/hooks/inject-plugin-root.sh",
+            "timeout": 10
+          }
+        ]
       }
     ]
   }

package/hooks/inject-plugin-root.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+# inject-plugin-root.sh — PostToolUse hook for Skill tool.
+# Fires after the presentation-generator skill loads and injects
+# PLUGIN_ROOT into additionalContext so the orchestrator never needs bash.
+
+INPUT=$(cat)
+SKILL=$(echo "$INPUT" | jq -r '.tool_input.skill // .tool_input.name // empty')
+
+# Only run for the presentation-generator skill
+case "$SKILL" in
+  *presentation-generator*) ;;
+  *) exit 0 ;;
+esac
+
+# Locate the plugin root
+PLUGIN_ROOT=$(find -L .claude ~/.claude \
+  -path "*/presentation-generator/scripts/outline_to_graph.py" \
+  2>/dev/null | head -1 | sed 's|/scripts/outline_to_graph.py||')
+
+[ -z "$PLUGIN_ROOT" ] && exit 0
+
+CONTEXT="PLUGIN_ROOT: $PLUGIN_ROOT"
+ESCAPED=$(echo "$CONTEXT" | python3 -c "import sys,json; print(json.dumps(sys.stdin.read()))")
+
+cat <<EOF
+{
+  "hookSpecificOutput": {
+    "hookEventName": "PostToolUse",
+    "additionalContext": $ESCAPED
+  }
+}
+EOF

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ztffn/presentation-generator-plugin",
-  "version": "1.5.0",
+  "version": "1.6.1",
   "description": "Claude Code plugin for generating graph-based presentations",
   "bin": {
     "presentation-generator-plugin": "bin/index.js"

package/scripts/outline_to_graph.py

@@ -93,6 +93,18 @@ def parse_drilldowns(text):
                 minor = _letter.get(dd_letter.group(1).upper(), 1)
                 title = dd_letter.group(2).strip()
                 drilldowns[current_parent].append((current_parent, minor, title))
+                continue
+
+            # Fallback C: "- **Title**" or "- **Title: desc**" or "- **Title — desc**"
+            # Agent wrote title-first bold instead of "- **N.M** Title:". Extract the
+            # title portion (before any colon, em-dash, or en-dash) and auto-number.
+            bold_title = re.match(r"-\s+\*\*([^*]+?)\*\*", line)
+            if bold_title:
+                raw = bold_title.group(1)
+                title = re.split(r"[:\u2014\u2013]", raw)[0].strip()
+                if title:
+                    minor = len(drilldowns[current_parent]) + 1
+                    drilldowns[current_parent].append((current_parent, minor, title))
 
     return drilldowns
 
@@ -131,6 +143,23 @@ def parse_content_blocks(text):
         flags=re.MULTILINE,
     )
 
+    # Normalize freeform drill-down headers:
+    #   ### Some Title
+    #   **(Drill-down under Spine Node N ...)**
+    # → ### SPINE N.M: Some Title  (M auto-incremented per parent)
+    _dd_minor: dict = {}
+    def _normalize_freeform_dd(m):
+        title = m.group(1).strip()
+        parent = int(m.group(2))
+        _dd_minor[parent] = _dd_minor.get(parent, 0) + 1
+        return f"### SPINE {parent}.{_dd_minor[parent]}: {title}"
+    text = re.sub(
+        r"^### (.+?)\n\*\*\(?Drill-down under Spine(?:\s+Node)?\s+(\d+)[^)]*\)?\*\*",
+        _normalize_freeform_dd,
+        text,
+        flags=re.MULTILINE,
+    )
+
     blocks = {}
     parts = re.split(r"(?=^### SPINE )", text, flags=re.MULTILINE)
 

package/skills/frameworks/SKILL.md

@@ -137,7 +137,8 @@ Frameworks define the spine structure. Drill-downs can use a different pattern:
 |---|---|---|
 | 5-10 minutes | 4-5 | 0-1 |
 | 15-20 minutes | 5-6 | 1-2 |
-| 30+ minutes | 6-7 | 2-3 |
+| 30-45 minutes | 7-9 | 2-4 |
+| 60-90 minutes | 8-10 | 4-7 |
 
 The spine is the minimum viable presentation. Drill-downs are optional depth the presenter can use or skip based on time and audience interest.
 

package/skills/graph-topology/SKILL.md

@@ -35,7 +35,7 @@ No drill-downs. Every slide is on the main horizontal path.
 
 ### 2. Spine with Deep Dives
 
-Main path has 4-7 spine nodes. Each can have 1-3 optional drill-down children below it.
+Main path has spine nodes scaled to presentation duration (see frameworks skill for the duration table). Each spine node can have optional drill-down children below it — longer presentations go deeper rather than wider.
 
 ```
 [Cover] → [Problem] → [Solution] → [Value] → [CTA]

package/skills/presentation-generator/SKILL.md

@@ -15,16 +15,24 @@ You are an orchestrator. Execute these 7 phases in order. Do not skip phases. Do
 
 ## Phase 1 — Setup
 
-Run this bash command to find the plugin:
+**Locate the plugin root.**
 
-```bash
-find -L .claude ~/.claude -path "*/presentation-generator/scripts/outline_to_graph.py" 2>/dev/null | head -1 | sed 's|/scripts/outline_to_graph.py||'
-```
+First, check if your context already contains a line beginning with `PLUGIN_ROOT:`
+(injected automatically by a hook when running in Claude Code). If found, use that
+value directly — you are done with this step.
+
+If not found (Cowork mode, or hook not active), use the **Read tool** to probe
+these paths in order — stop at the first one that returns file contents:
+
+1. `.claude/plugins/presentation-generator/scripts/outline_to_graph.py`
+2. `~/.claude/plugins/presentation-generator/scripts/outline_to_graph.py`
+
+PLUGIN_ROOT is everything before `/scripts/outline_to_graph.py` in the successful
+path (e.g. `.claude/plugins/presentation-generator`).
 
-The output is the PLUGIN_ROOT path (e.g. `.claude/plugins/presentation-generator`).
-If the output is empty, the plugin is not installed — stop and tell the user.
+If neither Read succeeds, stop and tell the user the plugin is not installed.
 
-Save the PLUGIN_ROOT value. You need it for every later phase.
+Save PLUGIN_ROOT — you need it for every later phase.
 
 Then extract from the user's message:
 - **TOPIC**: what is being presented
@@ -184,12 +192,36 @@ cp _temp/presentation-content-brief.json "presentations/{SLUG}/content-brief.jso
 cp _temp/presentation-outline.md "presentations/{SLUG}/outline.md"
 ```
 
-Then tell the user:
+Then upload the presentation automatically. Read the upload skill first:
+
+```
+Read: {PLUGIN_ROOT}/skills/upload-presentation/SKILL.md
+```
+
+The presentation JSON is at `presentations/{SLUG}/{SLUG}.json`. The `name` field must be at the JSON root — use the TOPIC from Phase 1:
+
+```bash
+jq --arg name "{TOPIC}" '. + {"name": $name}' "presentations/{SLUG}/{SLUG}.json" | \
+  curl -s -X POST "https://humagreenfield.netlify.app/api/presentations/upload" \
+    -H "Content-Type: application/json" \
+    -H "X-API-Key: TheSunIsFree" \
+    -d @-
+```
+
+On success, report the returned URLs to the user:
+
+```
+Presentation published:
+- Present: https://humagreenfield.netlify.app{presentUrl}
+- Edit:    https://humagreenfield.netlify.app{editUrl}
+```
+
+If the upload fails, fall back to manual import:
 
 ```
 Your presentation is at: presentations/{SLUG}/{SLUG}.json
 
-To import:
+To import manually:
 1. Open https://humagreenfield.netlify.app/present/plan
 2. New presentation → Import JSON
 3. Select the file above

package/skills/upload-presentation/SKILL.md

@@ -0,0 +1,155 @@
+---
+name: upload-presentation
+description: >
+  Upload a React Flow presentation graph JSON directly to the Huma Showcase platform via HTTP API.
+  Use when an agent has generated a presentation (e.g. from the presentation-generator skill) and
+  wants to publish it without manual browser interaction. Triggers on: "upload the presentation",
+  "publish the deck", "deploy the slides", "push to the showcase", "upload presentation-draft.json".
+---
+
+# Upload Presentation
+
+Publishes a presentation graph (nodes + edges) to the platform via a single HTTP call.
+Returns a live URL — no browser interaction needed.
+
+## Prerequisites
+
+- API key: `TheSunIsFree`
+- The JSON to upload must match the `GraphPayload` shape: `{ name, nodes[], edges[] }`.
+
+Pass the key via the `X-API-Key` header or `?key=TheSunIsFree` query param.
+
+## Upload Command
+
+```bash
+curl -s -X POST "https://humagreenfield.netlify.app/api/presentations/upload" \
+  -H "Content-Type: application/json" \
+  -H "X-API-Key: TheSunIsFree" \
+  -d @_temp/presentation-draft.json
+```
+
+Or with the key in the URL:
+
+```bash
+curl -s -X POST "https://humagreenfield.netlify.app/api/presentations/upload?key=TheSunIsFree" \
+  -H "Content-Type: application/json" \
+  -d @_temp/presentation-draft.json
+```
+
+## Request Body
+
+The file at `_temp/presentation-draft.json` must contain:
+
+```json
+{
+  "name": "Human-readable presentation title",
+  "description": "Optional one-line description",
+  "nodes": [ /* GraphNode[] — see node schema below */ ],
+  "edges": [ /* GraphEdge[] */ ]
+}
+```
+
+The API requires `name` at the JSON root. Supply it explicitly using the TOPIC from Phase 1:
+
+```bash
+jq --arg name "{TOPIC}" '. + {"name": $name}' "presentations/{SLUG}/{SLUG}.json" | \
+  curl -s -X POST "https://humagreenfield.netlify.app/api/presentations/upload" \
+    -H "Content-Type: application/json" \
+    -H "X-API-Key: TheSunIsFree" \
+    -d @-
+```
+
+If uploading a manually constructed payload, ensure `nodes` is present and non-empty (`[]` is accepted but produces a blank presentation).
+
+### Minimal node shape
+
+```json
+{
+  "id": "cover",
+  "type": "huma",
+  "position": { "x": 0, "y": 0 },
+  "data": {
+    "label": "Slide Title",
+    "content": "Optional markdown body",
+    "notes": "Optional speaker notes"
+  },
+  "style": { "width": 180, "height": 70 },
+  "measured": { "width": 180, "height": 70 }
+}
+```
+
+Full field reference: `.claude/skills/presentation-generator/system/node-schema.md`
+
+### Minimal edge shape
+
+```json
+{
+  "id": "cover-to-problem",
+  "source": "cover",
+  "target": "problem",
+  "sourceHandle": "s-right",
+  "targetHandle": "t-left"
+}
+```
+
+Handle pairs: `s-right/t-left` (→), `s-left/t-right` (←), `s-bottom/t-top` (↓), `s-top/t-bottom` (↑)
+
+## Response
+
+Success (`200`):
+
+```json
+{
+  "ok": true,
+  "slug": "q1-strategy",
+  "name": "Q1 Strategy",
+  "presentUrl": "/present/q1-strategy",
+  "editUrl": "/present/plan/q1-strategy"
+}
+```
+
+Prepend the site origin to get absolute URLs:
+- Present: `https://humagreenfield.netlify.app/present/q1-strategy`
+- Edit: `https://humagreenfield.netlify.app/present/plan/q1-strategy`
+
+Error responses use standard HTTP status codes with `{ "error": "..." }` body:
+
+| Status | `error` message | Fix |
+|--------|----------------|-----|
+| `400` | `"Invalid JSON body"` | Body is not valid JSON |
+| `400` | `"Body must be a JSON object"` | Wrap payload in `{}` |
+| `400` | `"name is required (string)"` | Add `"name"` field to the JSON root |
+| `400` | `"nodes is required (array)"` | Add `"nodes": [...]` to the JSON root |
+| `401` | `"Unauthorized"` | Check API key — must be `TheSunIsFree` |
+| `500` | `"Server misconfiguration: PRESENTATIONS_API_KEY is not set"` | Env var missing on server |
+| `500` | Storage error message | Blob write failed — retry or check Netlify status |
+
+## Slug Behavior
+
+The slug is derived from `name` via `slugify()` (lowercase, hyphens, max 64 chars).
+If the slug already exists, a numeric suffix is appended automatically (`-2`, `-3`, …).
+Bundled template slugs (`lyse-neo-pitch`, `presentation-demo`, `huma-partner-pitch`) are always protected.
+
+## Integration with Presentation Generator
+
+In Phase 7 of the `presentation-generator` pipeline, the output JSON is at `presentations/{SLUG}/{SLUG}.json`. Upload it with the TOPIC as the name:
+
+```bash
+jq --arg name "{TOPIC}" '. + {"name": $name}' "presentations/{SLUG}/{SLUG}.json" | \
+  curl -s -X POST "https://humagreenfield.netlify.app/api/presentations/upload" \
+    -H "Content-Type: application/json" \
+    -H "X-API-Key: TheSunIsFree" \
+    -d @-
+```
+
+## Local Dev
+
+Replace the Netlify URL with `http://localhost:3000`. The endpoint behaves identically — blobs
+fall back to `_temp/react-flow/presentations/` on the local filesystem.
+
+```bash
+curl -s -X POST "http://localhost:3000/api/presentations/upload" \
+  -H "Content-Type: application/json" \
+  -H "X-API-Key: TheSunIsFree" \
+  -d @_temp/presentation-draft.json
+```

package/DEFERRED-FIXES.md

@@ -1,38 +0,0 @@
-# Deferred Fixes
-
-Issues that require architectural changes and should be addressed in targeted, focused fixes rather than prompt-level patches.
-
----
-
-## v1.5.0 fixes ✓ resolved
-
-**Bug: Style agent always launched as `undefined`**
-`enforce-style-schema.sh` returned `updatedInput: { prompt: "..." }` which replaced
-the entire tool_input, dropping `subagent_type`. Fixed by merging the modified prompt
-into the original tool_input: `updatedInput: ($orig + {prompt: $prompt})`.
-
-**Bug: Content brief triggered presentation validator**
-Hook patterns `*_temp/presentation*.json` matched `_temp/presentation-content-brief.json`,
-causing the validator to reject it. Removed the broad arm — only `*_temp/presentation-draft.json`
-needs validation now.
-
-**Bug: Bare agent names caused first-attempt failures**
-SKILL.md used `presentation-content` etc; Task tool requires fully-qualified
-`presentation-generator:presentation-content`. Fixed in all three phases (2, 3, 6).
-
-**UX: Visual intent annotations leaking into slide content**
-Narrative agents sometimes write `**Visual intent:** bookend` inside `**Content:**` blocks.
-Added `**Visual intent:**` as a recognized field marker in `extract_fields()` so it acts
-as a boundary, preventing bleed into `data.content`. Documented in `outline-format/SKILL.md`.
-
-**UX: Intimidating inline scripts in permission prompts**
-Phase 1 multi-line if/elif replaced with a single `find` one-liner.
-Phase 5 Python heredoc extracted into `scripts/verify_node_count.py`.
-
----
-
-## Issue 3 — `topic` badge truncation ✓ resolved
-
-**Was:** Badge clips at ~180px wide when `topic` strings are long.
-
-**Resolution:** The root cause was slide titles being written at full descriptive length (10–15 words) rather than as sharp 6–10 word claims. Fixed by adding slide title length rules to `slide-content/SKILL.md` (narrative agent) and reinforcing the hard limit in the style agent's label rewriting rules. Short titles produce short `topic` strings naturally — no schema change or UI change required.