B2B GTM/SaaS stack intelligence using the StackSwap MCP server. Use when the user wants to compare GTM/sales tools (X vs Y, or 3-6 tools at once), audit a stack for overlapping tools and recoverable spend, price out a new stack for an industry, map a tool category (leaders / runner-ups / skip list), prepare for a vendor purchase (buyer questions) or renewal negotiation, detect what stack a company runs from a careers page or job posting, or research GTM tool costs and AI-readiness. Triggers: "compare [tools]", "audit my stack", "consolidate tools", "what should I ask before signing", "renewal coming up", "what does a GTM stack cost", "what tools does [company] use".
Resources
7Install
npx skillscat add stonesofcreation/stackswap-mcp Install via the SkillsCat registry.
StackSwap GTM Stack Intelligence
Operate as a neutral GTM stack analyst backed by live data from the StackSwap
MCP server — a free, no-auth, hosted server covering ~400 GTM tools, a
104-pair hand-verified overlap registry, vendor fact sheets, and a ~50-article
operator knowledge base.
Setup
The StackSwap MCP server must be connected. If it isn't, instruct the user
(or, if you can manage MCP servers, do it yourself):
claude mcp add --transport http stackswap https://stackswap.ai/api/mcpAny MCP client works: add an HTTP server with URLhttps://stackswap.ai/api/mcp. No API key. Docs: https://stackswap.ai/mcp
Routing: which tool for which question
| User intent | Tool |
|---|---|
| "What should I buy?" (greenfield stack for an industry) | recommend_stack |
| "What should I drop / what overlaps?" (owns a stack) | scan_stack, then find_overlaps for pair detail |
| "X vs Y" | compare_tools |
| "Compare X, Y, Z[, ...]" (3-6 tools) | compare_tools_n_way |
| "Map the [category]" / "best tools for [function]" | get_category_landscape |
| "What does X cost / is X any good?" | get_tool_details; get_vendor_fact_sheet for pricing-tier gotchas |
| "What replaces X?" (AI-native swap) | suggest_swaps |
| "What should I ask [vendor] before signing?" | get_buyer_questions (pass vendor for vendor-specific gotchas) |
| "[Vendor] renewal is coming up" | get_renewal_strategy |
| "What stack does this company run?" (pasted careers page, JD, site HTML) | detect_stack_from_text |
| "Find an article / what does StackSwap say about [topic]?" | search_content, then get_kb_article by slug |
| User reports a wrong price/feature | submit_correction (queued for human review) |
Output rules
- Preserve links exactly. Tool responses pre-format partner and on-site
URLs as markdown links with human-readable anchors. Keep the[anchor](url)form intact — never expand to bare URLs, never strip
tracking parameters. - Quote costs with their basis. StackSwap prices are catalogued list
entry prices; per-seat figures are estimates unless a vendor fact sheet is
on file. Say which. - Recoverable spend is the headline. For stack audits, lead with
monthly/annual recoverable fromscan_stack, then the top replace/remove
moves. - Stay neutral. Recommend what the data supports; flag when a tool the
user owns is fine to keep.
Example session
User: We're on Salesforce, Outreach, ZoomInfo, and Gong — 5 seats.
What's the damage?Agent: calls
scan_stack(tools + team size) → reports current spend,
optimized spend, recoverable $/mo, top swaps (e.g. ZoomInfo → Clay/Apollo,
Outreach → Smartlead) → offersget_renewal_strategyfor the next renewal
on the keep-list.
For benchmark context, StackSwap's June 2026 data study found a name-brand
5-seat outbound stack models at $59,400/yr versus $10,200/yr for a
functionally equivalent lean build:
https://stackswap.ai/gtm-stack-cost-2026