athal7/docs

Docs

Two tools depending on where the content lives:

| Signal | Tool | |---|---| | .google.com/document URL, Doc ID, user says "Google Doc" | gws | | *.atlassian.net/wiki URL, page ID, user says "Confluence" or "wiki" | confluence |

Run --help on each tool and subcommand first. This skill covers only what help text won't surface.

Google Docs

• gws docs documents get returns raw structural JSON, not readable text — use gws drive files export --mimeType text/plain to get readable content
• --output paths must be within the current directory — gws rejects absolute paths outside it. Use .scratch/ as the download target: mkdir -p .scratch && gws ... --output .scratch/filename.txt. .scratch/ is globally gitignored.
• The Docs API has no search — use gws drive files list with a CQL q param instead
• gws docs documents get output begins with a "Using keyring backend" line — strip non-JSON prefix lines before parsing
• Inserting real tables requires a multi-phase approach; see tables.md

Google Slides

• gws slides presentations create creates a blank presentation — use gws drive files copy on an existing themed presentation to inherit its master/theme instead
• Google's template gallery IDs are not accessible via the API (404) — copy from a presentation you already own that has the desired theme
• The default master (Simple Light) has no named layouts returned in masters[].layouts via the API, but predefined layouts (TITLE_AND_BODY, TITLE_ONLY, SECTION_HEADER, BIG_NUMBER, BLANK) work via createSlide.slideLayoutReference.predefinedLayout
• Use placeholders, not text boxes. Slides created with a predefined layout get TITLE and BODY placeholder shapes — insert text into those via insertText and the master handles font, size, and position. Manual text boxes (createShape: TEXT_BOX) bypass the master entirely and require you to hardcode all coordinates in EMU
• Canvas is 720×405pt (9144000×5143500 EMU) — pt → EMU conversion: pt * 12700. Placeholder positions are already set by the layout; you only need coordinates when placing decorative shapes
• Object IDs must be ≥5 characters and match [a-zA-Z0-9_][a-zA-Z0-9_\-:]* — no spaces
• foregroundColor in updateTextStyle takes {"opaqueColor": {"rgbColor": {...}}}, not {"solidFill": ...}. Page background and shape fills use {"solidFill": {"color": {"rgbColor": {...}}}}
• alignment in updateParagraphStyle uses START/CENTER/END, not LEFT/RIGHT
• batchUpdate request type is createSlide, not addSlide
• All slides in a batchUpdate must succeed or none are applied — validate object IDs and request shape before sending

Confluence

• confluence find matches by exact title — use confluence search for partial or full-text matches
• confluence search uses CQL — wrap multi-word terms in quotes
• Page IDs appear as pageId= in older URLs; newer Confluence Cloud URLs embed the ID in the path — confluence read accepts both forms
• Inline comment creation requires editor-generated highlight metadata not exposed by the public REST API — footer comments and replies work reliably, inline creation will 400