Tool Reference

Each tool matches one API endpoint, and your assistant only sees the tools that its key’s scopes allow. All rates are fractions between 0 and 1. The range input can be 7d, 30d or 90d, with 30d as the default, and every rate comes back with the previous window’s value and the change between them.

Me and workspace

Tool	Scope	Inputs	Returns
`whoami`	any	None	The workspace this key belongs to, the key’s name, scopes and expiry. The assistant calls this first when it isn’t sure which workspace it’s connected to.
`get_workspace`	`workspace:read`	None	Workspace name, website, logo, timezone, owner and member count.
`get_usage`	`workspace:read`	None	Plan, subscription state, credit balances per action type, plan limits, and prompts tracked against the cap.

AI visibility

Scope: visibility:read.

Tool	Inputs	Returns
`get_visibility_setup`	None	Brand name, domain and variants, tracked competitors and categories (with IDs), the collection schedule, and the AI platforms your plan collects from.
`get_visibility_overview`	`range`	Workspace-wide mention rate, citation rate and share of voice with deltas, plus daily trends. This is the best place to start for “how are we doing”.
`get_visibility_by_platform`	`range`	Snapshots, mention and citation rate per platform (ChatGPT, Gemini, Perplexity, Claude, Grok, Google AI Overview).
`get_visibility_by_category`	`range`	Mention and citation rates per prompt category, plus an uncategorised row.
`get_competitor_leaderboard`	`range`	Every competitor with mention rate, citation rate and share of voice, and where you rank on each. Each row carries `competitor_id` (null on your own row).
`get_competitor`	`competitor_id`, `range`	One competitor in depth: rates, prompt coverage, trend, citations by platform, top cited pages.
`get_competitor_pages`	`competitor_id`, `range`, `search?`, `url?`	Every page of theirs that AI cites, ordered by citations. With `url`, one page’s citation rate, share, trends and the prompts citing it.
`get_cited_sources`	`range`, `type` (`domain` \| `url`), `limit` (≤50)	The domains or URLs cited most across all your prompts, tagged as yours, a competitor’s or neither.
`list_cited_pages`	`range`, `search?`, `sort`, `order`, `page`, `per_page` (≤100)	Your own pages that AI cites: citations, citation rate, share and unique prompts.
`get_cited_page`	`url` (as listed, with or without `https://`), `range`	One of your pages: rates, share, trends and the prompts that cite it.
`list_prompts`	`range`, `search?`, `category_id?`, `sort`, `order`, `page`, `per_page` (≤100)	Tracked prompts with mention and citation rates. Returns prompt IDs. `sort` by `created_at`, `category`, `mention_rate` or `citation_rate`.
`get_prompt_analytics`	`prompt_id`, `range`	One prompt: rates and trends, per-platform results, competitors mentioned.
`get_prompt_answers`	`prompt_id`, `range`, `platform?`, `date?` (YYYY-MM-DD), `full`	The answers AI platforms gave, newest first, up to 200. With `full=true`, the answer text, citations, brand mention excerpts and detected competitors.

Content map

Scope: contentmap:read.

Tool	Inputs	Returns
`get_topic_matrix`	None	Every topic with your page count and each competitor’s, split by buyer stage, flagged `is_coverage_gap` (you have fewer than the top competitor) and `is_untapped` (you have none).
`get_sites_summary`	None	Your site and each competitor site: pages, analysed, pending, topics covered, buyer-stage mix. Returns entity IDs.
`list_site_pages`	`entity_id?`, `topic_id?`, `buyer_stage?` (`AWARENESS` \| `CONSIDERATION` \| `DECISION` \| `NA`), `status?`, `search?`, `page`, `per_page` (≤200)	Analysed pages across all sites, newest first. Combine `entity_id` and `topic_id` to see one competitor’s pages on one topic.

Content

Scope: content:read.

Tool	Inputs	Returns
`list_content`	`mode` (`IDEAS` \| `PLANNED` \| `PRODUCED`), `search?`, `status?`, `sources?`, `stages?`, `from?`, `to?`, `page`	The content library, 15 per page, without article bodies. Defaults to `PRODUCED` (the REST endpoint defaults to `IDEAS`). `IDEAS` filters by source and `stages` (`AWARENESS` \| `CONSIDERATION` \| `DECISION` \| `BALANCED`), and `PRODUCED` filters by status.
`get_article`	`content_id`	One item in full, including the markdown body, meta title and description, and where it was published.

Articles are also available as MCP resources at deepsmith://articles/{id}, so if your client lets you attach resources, you can pull an article straight into the conversation.

Deep IQ

Scope: iq:read.

Tool	Inputs	Returns
`get_brand_assets`	`type?` (`PRODUCT` \| `PERSONA` \| `VOICE` \| `CONTENT_TYPE` \| `VISUAL_GUIDELINE`), `search?`	Your IQ items with their generated content. Returns IQ IDs.
`get_brand_asset`	`iq_id`	One IQ item in full: content and metadata.

Prompts

These are ready-made analyses that combine the tools above. You can read Using the MCP server to see when each one is useful.

Prompt	Arguments	Needs
`weekly_visibility_briefing`	`range` (default `7d`)	`visibility:read`
`competitor_gap_analysis`	`competitor_id`, `range`	`visibility:read`, `contentmap:read`
`why_not_cited`	`prompt_id`, `range`	`visibility:read`, `contentmap:read`
`content_gap_plan`	`buyer_stage?`	`contentmap:read`, `content:read`
`answer_audit`	`range`, `platform?`	`visibility:read`

​Me and workspace

​AI visibility

​Content map

​Content

​Deep IQ

​Prompts