Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
Steve Jones avatar

Jack Tar Superpower Bridge

  • 3 repo stars
  • Updated July 24, 2026
  • SteveGJones/jack-tar-deckhand

Bridge pptx generation with narrative pre-briefing and post-hoc visual enrichment to combine structural decks with richer visuals like AI backgrounds and native charts.

About

jack-tar-superpower-bridge wraps the pptx document skill with a narrative pre-brief and post-hoc visual enrichment commands. It combines pptx's structural strength with jack-tar's visual capabilities such as AI backgrounds, element images, editable SmartArt, and native charts. The skill helps produce richer, better-designed presentations.

  • Narrative pre-brief command
  • Post-hoc deck enrichment
  • AI backgrounds + element images
  • Editable SmartArt and charts

Jack Tar Superpower Bridge by the numbers

  • Data as of Jul 25, 2026 (Skillselion catalog sync)
/plugin marketplace add SteveGJones/jack-tar-deckhand
/plugin install jack-tar-superpower-bridge@jack-tar

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
repo stars3
Last updatedJuly 24, 2026
RepositorySteveGJones/jack-tar-deckhand

What it does

Bridge pptx generation with narrative pre-briefing and post-hoc visual enrichment to combine structural decks with richer visuals like AI backgrounds and native charts.

README.md

Jack-Tar — Deckhand

"A Jack-Tar never lets the deck go untended."

AI-powered skills and agents for building conference-quality presentation decks with Claude Code and Claude Desktop.


Vision

The upstream /pptx skill produces solid slide decks, but conference-quality presentations demand more: bespoke hero imagery, data-driven infographics, speaker-ready layouts, and a visual identity that holds up on a 40-foot projector screen in front of 2,000 people. This project closes that gap.

Jack-Tar is a coordinated suite of Claude skills and orchestration agents that combine image generation models, layout intelligence, and content-authoring tools into a single end-to-end pipeline. It offers two entry points: the Superpower Bridge route (the default, which collaborates with /pptx) and the Direct route (full Jack-Tar pipeline from a brief, no /pptx involvement). A speaker describes their talk — through either route — and the system delivers a polished, stage-ready .pptx.


Problem Statement

Building a conference deck today — even with AI assistance — still involves a fragmented workflow:

  1. Draft an outline in one tool.
  2. Generate images in another (Midjourney, DALL·E, Flux, etc.).
  3. Manually place and resize assets in PowerPoint or Google Slides.
  4. Iterate on layout, colour, and typography by hand.
  5. Write speaker notes separately.
  6. Run a visual QA pass yourself.

Each handoff loses context. The image generator doesn't know your slide dimensions. The layout tool doesn't know your narrative arc. Nobody enforces brand consistency across 30 slides. Jack-Tar eliminates these seams by keeping the entire pipeline inside Claude's skill and agent framework, where every component shares context.


Two routes at a glance

Jack-Tar offers two entry points: the Superpower Bridge route (default, collaborates with /pptx) and the Direct route (full Jack-Tar pipeline from a brief).

Both routes support 1K / 2K / 4K cloud-rendered visuals via the jack-tar-cloud plugin (see EPIC #58).


Choosing your route

                   Have you already run /pptx and want to fix the deck?
                                         │
                  ┌──────────────────────┴──────────────────────┐
                 Yes                                            No
                  │                                              │
                  ▼                                              ▼
      Bridge route, review-first              Want to keep using /pptx?
      ──────────────────────────                          │
      /enrich-deck output.pptx           ┌────────────────┴────────────────┐
                                        Yes                               No
                                          │                               │
                                          ▼                               ▼
                            Bridge route (default)              Direct route
                            ──────────────────────              ────────────
                            /bridge-brief →                /jack-tar-deckhand:
                            /pptx →                          deck-conductor
                            /enrich-deck

1. Bridge from the start (default)

Speaker invokes /bridge-brief first; uses the brief to drive /pptx; then /enrich-deck layers visuals onto the resulting deck. Speaker collaborates with the bridge throughout — visuals are designed for, not retrofitted onto, the slides.

2. Bridge after a stale /pptx run

Speaker has already run /pptx and isn't satisfied with the deck. The bridge takes a review-first approach: assesses what's there, identifies what's salvageable vs what needs redoing, and collaborates with the speaker on whether to enrich in place or rebuild the brief and start over. This is how the bridge handles "rescue" workflows.

3. Jack-Tar direct route

Speaker dispatches the deck-conductor agent for the full Jack-Tar pipeline — brand profile, narrative architect, strategy map, full QA, no /pptx involvement at all. Best for speakers who want the full Jack-Tar treatment from a brief, not retrofitting onto an existing deck.


Quick Start

Bridge route (default)

/bridge-brief "your topic"
/pptx ...                   # use the brief to drive the upstream /pptx skill
/enrich-deck output.pptx

Direct route

Dispatch the deck-conductor agent with your topic to orchestrate a full deck build (it is a conversational agent, not a slash command):

Use the deck-conductor agent to build a deck on "your topic"


Architecture Overview

Jack-Tar is a 5-plugin Claude Code marketplace. The two pipelines share a set of engine plugins (image generation, SmartArt) while differing in their entry point and orchestration layer.

Direct route

┌─────────────────────────────────────────────────────────┐
│                   Orchestration Layer                    │
│          deck-conductor agent (top-level)                │
│   Receives talk brief → coordinates all skills → .pptx  │
└────────────┬──────────────┬──────────────┬──────────────┘
             │              │              │
     ┌───────▼──────┐ ┌────▼─────┐ ┌──────▼───────┐
     │  Content      │ │  Visual   │ │  Assembly &  │
     │  Skills       │ │  Skills   │ │  QA Skills   │
     │              │ │           │ │              │
     │ • narrative   │ │ • imagegen│ │ • layout     │
     │ • speaker-    │ │ • iconset │ │ • brand-qa   │
     │   notes       │ │ • palette │ │ • slide-qa   │
     │ • outline     │ │ • chart   │ │ • pptx-build │
     └──────────────┘ └───────────┘ └──────────────┘

Diagram labels are functional roles; mapped skill names are listed in §8 below.

Bridge route

/pptx deck ──► /bridge-brief or /enrich-deck ──► enriched .pptx
                     (collaborates with deckhand skills)

Project Components

8.1 Direct-route components

Skill Purpose
deck-conductor Orchestration agent — accepts talk brief, coordinates all skills
narrative-architect Transforms brief into structured slide outline
imagegen-bridge Routes image generation to available engine plugins
slide-stylist Owns visual design system: palette, typography, layout templates
speaker-notes-writer Generates timed, cued speaker notes per slide
chart-renderer Generates publication-quality charts as slide-ready images
deck-assembler Final composition — produces the .pptx
deck-qa Automated QA: overlap, contrast, margins, text overflow, consistency

8.2 Bridge-route components

Skill Purpose
/bridge-brief Plan a talk and produce a brief that drives the upstream /pptx skill
/enrich-deck Review-first enrichment of an existing .pptx — assess salvageability, collaborate on enrich-in-place vs rebuild
shared engines jack-tar-cloud, jack-tar-ollama, jack-tar-msft-smartart, jack-tar-custom-smartart

Plugin Marketplace

Plugin Purpose Route(s) served
jack-tar-deckhand Full presentation pipeline orchestrator Direct
jack-tar-superpower-bridge Bridge between /pptx and Jack-Tar engine plugins Bridge
jack-tar-ollama Local AI image generation (draft tier, free) Both
jack-tar-cloud Cloud AI image generation (production tier, 1K/2K/4K) Both
jack-tar-msft-smartart Editable PowerPoint SmartArt (29 layouts) Both
jack-tar-custom-smartart Data viz and custom graphics (SVG, Mermaid, Vega) Both

Plugin documentation: plugins/jack-tar-deckhand/CLAUDE.md · plugins/jack-tar-superpower-bridge/CLAUDE.md


Shared Data Contracts

The following contracts apply to the direct route pipeline. All skills communicate through well-defined JSON interfaces:

TalkBrief — Input from the user describing their presentation needs. Fields include topic, audience, duration_minutes, tone, key_takeaways, branding (optional logo, corporate colours), and preferences (e.g., "minimalist," "data-heavy," "image-rich").

DeckContext — The shared state object that the conductor passes between skills. Accumulates outputs as the pipeline progresses: starts with the TalkBrief, gains the StyleGuide after the stylist runs, gains the SlideOutline after the narrative architect runs, gains the ImageManifest after image generation, and so on.

SlideOutline — Array of slide objects, each with slide_number, slide_type, headline, body_points, narrative_beat, visual_direction, data (optional, for chart slides), and speaker_notes.

StyleGuide — Palette, font pairings, spacing scale, layout template assignments per slide type, and brand override tokens.

ImageManifest — Array of generated image records: slide_number, file_path, dimensions, alt_text, source_prompt, model_used, placement_zone.

QAReport — Array of findings: slide_number, severity (error, warning, info), category, description, suggested_fix.

These contracts are formally defined as JSON schemas in src/schemas/ and validated at each handoff point.


Development Approach

This project follows a high-ceremony, quality-first SDLC:

Skill-by-skill delivery. Each skill is developed, tested, and stabilised independently before integration. The delivery order follows the dependency graph: slide-stylist and narrative-architect first (no upstream dependencies), then imagegen-bridge and chart-renderer (depend on style guide), then deck-assembler (depends on all content and visual outputs), then deck-qa (depends on assembled deck), and finally deck-conductor (orchestrates everything).

Eval-driven development. Every skill gets a set of test prompts (evals/prompts/) and assertion-based benchmarks (evals/evals.json) before it ships. The skill-creator framework's eval loop (draft → test → review → improve → repeat) is the standard workflow for every component.

Visual QA as a gate. No skill is considered complete until its outputs have passed both automated QA checks and human visual review. For image generation and layout skills, this means rendering to images and inspecting every slide.

Contract-first integration. The JSON data contracts above are defined before implementation begins. Skills are developed against the contracts, and integration testing validates that each skill produces and consumes the correct shapes.

Documentation as a deliverable. Every skill ships with a complete SKILL.md, reference docs for any non-trivial logic, and inline comments in scripts. The docs/ directory is maintained alongside the code, not after.


Compatibility

Claude Code — Full support. The conductor agent can orchestrate multi-skill workflows using subagents, and the eval framework runs natively.

Claude Desktop — Supported for single-skill invocations and for the full pipeline when skills are installed locally. The conductor adapts its execution strategy (sequential rather than parallel) based on the environment.

Image generation backends — Require API keys configured as environment variables or MCP server connections. The imagegen-bridge skill documents the setup for each supported backend.


Status

  • BSA v1.4.0 (33 services, 6 AI personas, 60 interactions)
  • 1010 monorepo + 33 cross-plugin integration tests passing
  • Bridge route shipped through v0.1.0 → v0.2.0 (jack-tar-superpower-bridge plugin)
  • EPIC #58 (1K / 2K / 4K resolution capability) — 1K available via jack-tar-cloud; 2K / 4K wiring in flight (#59 / #60 / #61)

Roadmap

Phase Milestone Key Deliverables Status
0 — Foundation Data contracts & project scaffolding JSON schemas for all contracts, project directory structure, eval prompt library Done
1 — Content Core narrative-architect + speaker-notes-writer Working outline generation, speaker notes, eval results Done
2 — Visual Identity slide-stylist + chart-renderer Palette derivation, layout templates, chart rendering pipeline Done
3 — Image Pipeline imagegen-bridge Prompt engineering layer, backends, batch generation with selection Done
4 — Assembly deck-assembler End-to-end .pptx build from all upstream outputs Done
5 — Quality Gate deck-qa Automated QA checks, contrast/overlap/margin enforcement Done
6 — Orchestration deck-conductor Full pipeline from talk brief to finished deck Done
7 — Polish & Eval Integration testing at scale 1010 monorepo + 33 cross-plugin tests, SmartArt intelligent graphics, pptx_native engine Done
EPIC #58 Resolution capability 1K / 2K / 4K cloud-rendered visuals via jack-tar-cloud; available both routes In progress
Bridge maturation jack-tar-superpower-bridge v1.0 Full bridge plugin release, /bridge-brief and /enrich-deck stable Planned

License

TBD — to be determined.

Related skills

Office & Documentsworkflownotes

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.