Slides Cog

slides-cog 1.0.16 - Added explicit requirements to metadata: now specifies required binaries (python3) and environment variable (CELLCOG_API_KEY). - No changes to functionality or user-facing features.

slides-cog 1.0.15 - Updated usage instructions for agent support: clarified "All agents except OpenClaw" section. - Streamlined description for conciseness and clarity. - Minor wording and formatting improvements throughout documentation. - No functional changes to code or features.

- Updated skill description to highlight AI-driven deep research, SlideCog's #1 ranking on DeepResearch Bench (Apr 2026), and improved slide design features. - Added code example for initializing the CellCog client in Cursor, Claude Code, and Codex environments. - Clarified that deep research fills in content from minimal prompts. - No changes to features or usage—documentation improvements only.

slides-cog 1.0.13 Changelog - SKILL.md updated for brevity, clarity, and easier onboarding. - Simplified description, removing excessive marketing claims and outdated benchmarks. - Added a clearer "How to Use" section with concise usage instructions. - Trimmed repetitive statements; focused on practical guidance for formats, workflows, and suitable presentation types. - Retained full example prompts and detailed feature tables.

- Expanded and improved documentation in SKILL.md with detailed descriptions of supported presentation types, features, and output formats. - Clarified and emphasized differences between PDF and PPTX outputs, with dedicated sections on when and how to use each. - Added extensive example prompts and use cases for pitch decks, business, sales, educational, event, and image slideshow presentations. - Included new explanation of presentation features, slide types, and recommended chat modes for common scenarios. - Updated tone and messaging for greater clarity, professionalism, and helpfulness.

slides-cog 1.0.11 - Major SKILL.md rewrite for clarity and conciseness. - Simplified descriptions of features, formats, and use cases. - Clearer emphasis on PDF as default and PPTX as optional. - Removed redundancies and pared down examples and comparisons. - Added section linking to related skills for document and spreadsheet generation.

- Updated SKILL.md to add separate quick start instructions for OpenClaw and other agents, clarifying usage of notify_session_key for OpenClaw. - General SDK usage guidance improved; references to timeouts and file handling remain. - No changes to skill features, outputs, or dependencies.

- Updated the Quick Start section in SKILL.md for simpler, more direct usage instructions. - Added a pointer to the main cellcog skill for detailed SDK API documentation. - Replaced the detailed code example with a shorter Quick Start snippet referencing cellcog for full options. - No functional changes to the skill; documentation improvements only.

- Updated DeepResearch Bench ranking from February 2026 to April 2026 in the description and feature list. - No changes to code or functionality; documentation update only.

- Expanded and clarified the skill description for broader audiences. - Added metadata including supported operating systems and a homepage link. - Streamlined introductory and prerequisite sections for easier onboarding. - No changes to features or functionality—documentation improvements only.

- Added first-class support for native PPTX (PowerPoint) presentations; now generates editable slides on request. - Updated default output format guidance: PDF is generated when format is unspecified, PPTX on explicit request. - Clarified differences and recommendations for PDF vs. PPTX, including use cases. - Improved documentation for output formats, making it easier to understand when to use each. - Updated skill description and feature summary for greater clarity.

- Added explicit author and dependencies fields to metadata for clarity and compatibility. - Updated prerequisites section to reference the cellcog skill using consistent formatting. - No changes to functionality; documentation and metadata improvements only.

- Renamed the skill from "think-cog" to "slides-cog" and updated description to focus on slide and presentation generation. - Rewrote all documentation to center around PDF-first slide and presentation creation, highlighting content quality and visual design. - Added detailed sections describing supported presentation types: pitch decks, business & sales decks, educational slides, event presentations, and image slideshows. - Clarified default output is PDF, with explicit instructions for requesting PPTX/DOCX if needed (with noted quality tradeoffs). - Provided new usage examples and prompt patterns specific to presentation workflows. - Updated guidance on recommended agent modes for standard decks vs. high-stakes narrative slides.

- Skill renamed from slides-cog (presentation generation) to think-cog, focused on complex problem-solving and iterative idea exploration. - Purpose shifted: now centered on collaborative reasoning, brainstorming, and iterative strategy rather than creating presentations. - Updated documentation to highlight use cases such as architecture decisions, business strategy, creative direction, debugging, and structured decision-making. - Emphasizes conversational workflows (not fire-and-forget), with back-and-forth dialogue designed to help users figure out solutions through iteration. - Added guidance and tips for effective thinking sessions, contrasting think-cog with other CellCog "fire-and-forget" skills.

- Slides Cog now defaults to PDF output for all presentations; PDF is presented as the only standard format. - Strongly de-emphasized and demoted PPTX/DOCX support: users must now explicitly request these formats, and are warned of substantial quality loss. - Updated documentation to clarify that PDF is the preferred and future-proof format for AI-generated slides, with no format-prompting or ambiguity. - Emphasized shift in AI presentation creation: finished, polished results (not editable drafts) are the expected output paradigm. - Preserved support for a wide range of presentation types—business, education, events, image slideshows—while clarifying output expectations.

**Improved format recommendations and clarified best practices for CellCog-powered presentations.** - Added strong recommendation to use PDF format for best results; clarified quality differences between PDF, HTML, and PPTX output. - Updated instructions and quickstart to reflect the new create_chat pattern and recommended chat modes for different presentation scenarios. - Added guidance to recommend PDF to users first, and to explain limitations of PPTX/DOCX for AI-generated content. - Clarified when to use "agent" vs. "agent team" chat modes, simplifying most use cases to "agent". - Minor documentation improvements and new metadata (emoji) for the skill.

slides-cog v1.0.0 - Initial release of Slides Cog skill for AI-powered presentation generation. - Supports creating pitch decks, business, sales, educational, event presentations, and image slideshows. - Outputs presentations in PPTX, PDF, and Interactive HTML formats. - Requires CellCog skill for setup; uses agent team mode for presentations. - Includes comprehensive usage patterns, prompt tips, and presentation customization options.