jack-tar-deckhand

Produce a labeled figure with PERFECT text — render a label-free image locally (or take an external image) then overlay leader lines and typeset labels programmatically from vision-derived anchor coordinates. Text is correct by construction; only pointer placement is reviewed.

Ask which image models fit a set of tasks — get evidence-based recommendations with per-image costs, which external services to pay for, and which local models to install (with disk/RAM/licence guidance). Reads the live model catalog and probes what is actually available on this machine.

Check mflux (MLX) availability, per-family runtime presence, cached weights, and report readiness status.

Top-level image orchestrator. Routes all slide image generation to the appropriate skill (jack-tar-ollama:image, jack-tar-ollama:icon, jack-tar-ollama:pattern, jack-tar-ollama:diagram, jack-tar-cloud:image, jack-tar-cloud:icon, render_chart). Produces ImageManifest and ChartManifest. Also reads strategy-map.json to determine per-slide rendering approach (full_render, backdrop_render, composed).

Meta-verify — discover all jack-tar engine plugins, call each verify, report aggregate pipeline capability and discipline-hook readiness.

Generate an image locally on Apple Silicon via the mflux CLI (MLX). Flag-compatible with jack-tar-ollama's /image — a $0 local tier, no API keys needed.

Check which cloud image providers are configured and report readiness status.

Fetch the latest model catalog (model names, capabilities, pricing) from the repository and install it locally after an operator-gated price review — no plugin release required.

Generate images using Google Gemini/Imagen API. Requires GOOGLE_API_KEY (Gemini Developer API) or GOOGLE_APPLICATION_CREDENTIALS (Vertex).

Smart router — generates an image using the best available cloud provider. Tries providers in priority order: FAL (FLUX), OpenAI (GPT Image), Google (Imagen). Requires at least one provider configured.

Transform a TalkBrief into a structured SlideOutline through collaborative arc selection and autonomous outline generation.

Phase 3 of the superpower bridge — analyse a /pptx-produced .pptx, present enrichment options, apply selected enrichments transactionally to a copy, and deliver presentation-enriched.pptx with a structured enrichment-report.md. Never modifies the original. Reuses image-reviewer and prompt-engineer agents and dispatches the new enrichment-cohesion-reviewer for deck-level visual review.

Generate vector icons (SVG) using Recraft V4 API (direct or via FAL.ai). Accepts a text prompt with optional brand colors, generates native SVG output, and saves to a path. Requires RECRAFT_API_KEY (Recraft direct) or FAL_KEY (FAL.ai route).

Assemble a .pptx presentation from all DeckContext contracts (SlideOutline, StyleGuide, ImageManifest, ChartManifest, SpeakerNotes) via PptxGenJS.

Run 25 automated anti-pattern checks on an assembled .pptx and produce a QAReport with verdict and per-slide findings.

Generate timed, transition-cued speaker notes calibrated to talk duration and audience profile, with lightweight preference gathering.

Generate an editable SmartArt carrier .pptx from a spec. Accepts a SmartArt spec (graphic_type, items or tree, layout_id), renders the carrier file, and returns the path.

Classify each slide's rendering strategy (full_render, full_bleed, background, backdrop, pragmatic_composition, composed) and generate image prompts via the prompt-engineer agent.

Single-slide critique-driven refinement for academic_figure slides. Three modes — auto (Critic-driven, for flow diagrams), enumerate (operator-checklist, for completeness artefacts), draft (hybrid). Wraps paperbanana's --continue-run via subprocess with manifest update + failsafe rollback.

Phase 1 of the superpower bridge — collaborative narrative pre-brief for the /pptx superpower. Dispatches the Narrative Brief Architect persona to propose arcs, capture communication intent, and produce a creative-brief.md that guides /pptx to build a well-structured deck with placeholder markers where visual enrichment will live.

Generate raster images using Recraft V4 (1K standard, 2K Pro, 4K via Creative Upscale). Brand-color-fidelity provider — best for slides where exact hex compliance matters more than photorealistic detail. Requires RECRAFT_API_KEY (direct) or FAL_KEY (via FAL).

Generate images using local Ollama models with optional iterative refinement. Accepts a text prompt or prompt file, generates via Ollama, optionally reviews and refines using vision, saves to a specified path, and returns the file path.

Check plugin readiness — verifies python-pptx, lxml, esprima are importable, that the jack-tar-msft-smartart plugin is locatable, and that the jack-tar-deckhand plugin's imagegen-bridge skill is reachable. Reports a STATUS line and a per-dependency table.

Generate images using FAL.ai FLUX API. Requires FAL_KEY environment variable.

Generate images using OpenAI GPT Image API. Requires OPENAI_API_KEY environment variable.

Check Mermaid CLI, Vega CLI, matplotlib, cairosvg availability and report readiness.

Create, load, or update a reusable BrandProfile from brand assets (PDF guidelines, .pptx template, logo, manual input). Persists profiles for cross-deck reuse.

Derive a complete StyleGuide through collaborative design exploration with the Speaker, working within brand constraints when a BrandProfile exists.

Transforms approved slide content into engine-specific structured data for SmartArt rendering. Reads SmartArtRecommendations, extracts data from SlideOutline body_points, validates against target engine schemas, handles overflow. Writes smartart-spec.json.

Orchestrates the SmartArt type selection negotiation. Dispatches the smartart-selector agent to recommend graphic types and enrichment tiers, then runs approval/rejection loop with the narrative-architect. Writes smartart-recommendations.json.

Render a data chart (bar, line, area, pie, donut, scatter) to PNG using matplotlib. Accepts chart type and data, produces a rasterised PNG.

Render a SmartArt graphic to PNG using SVG, Mermaid, Vega-Lite, or matplotlib. Accepts a structured spec and produces a rasterised PNG.

Smart router for vector icon generation. Tries Recraft V4 SVG first, falls back to OpenAI if unavailable.

List all available SmartArt layouts with their capacities, data shapes, and recommended use cases.

Post-process an assembled deck — graft SmartArt from carrier files into the host .pptx. Reads smartart-spec.json from the deck directory to find carrier files and target slides.

Generate technical diagrams — flowcharts, architecture diagrams, system diagrams, and process flows. Uses FLUX model for better text rendering and spatial accuracy.

Generate application icons, favicons, and document markers. Produces clean, recognizable icons optimized for small sizes with optional multi-size output.

Generate seamless textures and repeating patterns for backgrounds, materials, and design assets.

Create visually rich PowerPoint presentations with AI-generated images, icons, diagrams, and patterns. Orchestrates the pptx skill with image generation skills to produce slides that aren't just text-on-white.

Check python-pptx, lxml, layout fixtures, and test carrier build capability.