Automate Google Slides Presentations

This skill provides lightweight Google Slides integration with standalone OAuth authentication and full read/write access - no MCP server required. Authentication is handled by scripts/auth.py, and all presentation operations run through scripts/slides.py, which auto-authenticates on first use if you're not already logged in.

First-time setup opens a browser to authenticate:

python scripts/auth.py login

Read operations include extracting all text content from a presentation (title, text from shapes and text boxes, and table cell data), finding presentations by search query with an optional result limit, and fetching metadata like title, slide count, slide object IDs, page size, and whether the presentation has masters or layouts. Write operations include creating a new empty presentation, adding a blank or specifically-laid-out slide (optionally at a given position), finding and replacing text across all slides (with optional case matching), deleting a slide by its object ID, and an advanced batch-update command that accepts raw Slides API request JSON for formatting, shapes, or images.

Nine slide layouts are available for add-slide --layout: BLANK, TITLE, TITLE_AND_BODY, TITLE_AND_TWO_COLUMNS, TITLE_ONLY, SECTION_HEADER, ONE_COLUMN_TEXT, MAIN_POINT, and BIG_NUMBER. Presentations can be referenced by either a bare ID or a full Google Slides URL, which the scripts parse automatically. Auth tokens are stored securely in the system keyring - Keychain on macOS, Windows Credential Locker on Windows, and Secret Service API (GNOME Keyring, KDE Wallet, etc.) on Linux - under the service name google-slides-skill-oauth, and expired tokens refresh automatically.

Use it when you need to create, inspect, or modify Google Slides presentations from local automation - reading slide text, adding or removing slides, or batch-updating content - without standing up an MCP server. It requires a Google Workspace account; personal Gmail accounts are not supported.

Use this skill only when the task clearly matches the scope described above - it is not a substitute for environment-specific validation, testing, or expert review, and you should stop and ask for clarification if required inputs, permissions, or success criteria are missing.

Inputs are a presentation ID or URL, plus command-specific arguments (search query, layout name, position index, find/replace text, or raw batch-update JSON). Outputs are extracted text, JSON search results with presentation IDs and modified times, or JSON metadata (title, slide count, page dimensions, layout flags).

Automation builders on Google Workspace who need scripted, local read/write access to Slides presentations - reporting pipelines, bulk content updates, or presentation generation - without deploying an MCP server.