Work with GitHub through the `gh` CLI — read and file issues, open and review pull requests, watch CI runs, inspect diffs and releases, and call the REST/GraphQL API. Use whenever a task mentions a GitHub issue, PR, repository, workflow run, or a github.com…
Work with Gmail, Google Calendar, Drive, Contacts, Sheets, and Docs through the `gog` CLI. Use whenever a task involves the operator’s email, calendar, Drive files, contacts, or a Google Sheet or Doc.
Read and write Notion through its REST API — search pages and databases, read a page as text, append blocks, create pages, and query a database with filters. Use whenever a task mentions Notion, a notion.so URL, or a Notion page or database.
Read and write Microsoft Word (.docx) and Excel (.xlsx) files — extract a document as text, turn a spreadsheet into JSON, and generate formatted documents and workbooks. Use whenever a task involves a .docx or .xlsx file, a Word document, or an Excel workbook.
Extract the real requirements — API keys and command-line prerequisites — from an installed skill bundle. Use when auditing a skill bundle for what it needs to run, and whenever a porting/mapping turn asks for a requirements proposal as JSON (`bakin skills…
Read text out of images, screenshots, and scanned PDFs with local Apple Vision OCR. Use when a task references an image or scanned document whose text needs to be read or quoted exactly.
Drive a real Chrome browser via CDP — open pages, read rendered content, run JavaScript, screenshot, extract readable article text. Use when a task needs what a page ACTUALLY renders (JS-heavy sites), a screenshot, or in-page interaction.
Transcribe audio or video files to timestamped text, fully locally. Use when a task references an audio/video file that needs its content read.