Stigg says its fastest-growing evaluator is now a coding agent that reads the docs and calls the MCP server, and the new Head of Growth owns "the MCP Registry" and "agent-readable docs and llms.txt". Today a search of the official MCP Registry for "stigg" returns nothing (Stripe has 29 entries), and the llms.txt that every docs page tells agents to read first still introduces Stigg as "a subscription and entitlement management platform for SaaS companies", without naming the MCP server, the CLI or npx @stigg/terminal. An agent asked to "add AI credits to my app" gets an older, narrower picture of Stigg than the homepage gives.
First action
Submit the MCP server to the official MCP Registry today.
How to measure it
Agent-sourced signups (terminal, MCP, Stripe Projects)
- Now
- –
- Target
- 2x
- By
- 90 days
Steps
List the server
Submit mcp.stigg.io with name, description, auth method and docs link to the official MCP Registry.
Head of Growth, Self-Serve & Agents 1 day
Rewrite the summary
New first paragraph of docs llms.txt: usage runtime for AI products, credits, entitlements, metering, governance.
Senior Product Manager 2 hours
Add an agent start block
npx @stigg/terminal, MCP URL, CLI, Build plan limits and where to get an API key, at the top of llms.txt.
Senior Product Manager 2 hours
Promote key pages
Link MCP Server, Agent Skills and the four AI-agent guides directly from the top-level index.
Senior Product Manager 1 hour
Run the agent test
Ask three coding assistants to add credits to a sample app; record the tool picked and the install path.
Growth Engineer (to be hired) 2 days
Fix what agents miss
Edit docs pages the agents misread or skipped.
Senior Product Manager 1 week
Repeat monthly
Rerun the same prompts and track the pick rate.
Head of Growth, Self-Serve & Agents monthly
Why it works
Publish mcp.stigg.io to the official MCP Registry (and the client directories that read from it) with a clear description: credits, entitlements, metering and budgets for AI products. Rewrite the first lines of docs llms.txt to match "the usage runtime for AI products", add a short "Start here for agents" block (npx @stigg/terminal, the MCP server, CLI, the free Build plan and its limits), and link the MCP Server, Agent Skills and AI-agent guides at the top level. Then test the result: ask three popular coding assistants to add credits to a sample app and record whether they pick Stigg and how they install it.
Agents choose tools from what registries and llms.txt files tell them. Getting listed and describing the current product costs a day and puts Stigg in front of the evaluator it says is growing fastest.
Ready to use
ChecklistAgent discovery launch list
[ ] mcp.stigg.io listed in the official MCP Registry [ ] docs llms.txt first line: "the usage runtime for AI products" [ ] "Start here for agents" block: npx @stigg/terminal, MCP URL, CLI [ ] Build plan limits stated (10,000 entities, 5M events) [ ] MCP Server and Agent Skills linked at top level [ ] Monthly three-assistant test with results logged
Risks and how to handle them
Registry listing alone changes little
Pair it with the llms.txt rewrite and measure installs from each.
Beta server confuses agents
State "public beta" and the supported clients in the listing.
Evidence
- HiringSame post: "more and more, the 'user' evaluating Stigg is not a person. It is a coding agent reading our docs, calling our MCP server and wiring up entitlements". The role owns "AI search (AEO / GEO), the MCP Registry and agent marketplaces, agent-readable docs and llms.txt, CLI and API-first signup, and agent-native payments (x402, Stripe's Machine Payments Protocol and whatever comes next)" and must "Measure how much of our growth comes from agents, then grow it"; get a developer "from signup to a first live entitlement check in minutes"; "Define product-qualified leads, run an inside sales motion"; direct agencies for SEO/AEO, paid and content.Source
- Your docsdocs.stigg.io/llms.txt (checked October 23, 2026), the index coding agents are pointed to from every docs page, opens: "Stigg is a subscription and entitlement management platform for SaaS companies." Its guides list covers credits, AI-agent metering and paywalls, but it names no MCP server, CLI or the npx @stigg/terminal installer; the MCP Server and Agent Skills pages appear only one level down, in the 278-page documentation.txt.Source
- MCP Registry checkThe official MCP Registry (registry.modelcontextprotocol.io) returned 0 servers for a search on "stigg" on October 23, 2026; the same search for "stripe" returned 29. Stigg's MCP server is in public beta at https://mcp.stigg.io (with an @stigg/typescript-mcp local bridge) and authenticates with an environment API key copied from the Stigg app; scoped keys are available on the Scale plan.Source
- Your websiteHomepage: title "Stigg - The Usage Runtime for AI Products"; H1 "Every AI request is a spend decision. Make it in milliseconds."; two calls to action, "Install with AI" (npx @stigg/terminal) and "Talk to a Human"; claims "p99 < 10ms" entitlement checks and "1M+ events/sec". www.stigg.io/llms.txt returns 404, so the site's product, pricing and comparison pages have no agent-readable summary.Source
- HiringHead of Growth, Self-Serve & Agents (Tel Aviv, posted October 6, 2026): "Since Stigg 2.0, self-serve signups from AI startups have roughly doubled, with no dedicated team behind them. Winning net-new AI logos is now the most important number in the company." The hire builds "Stigg's first product-led channel as a self-sufficient squad, with your own engineer, your own inside sales and your own budget" and owns "the number from first visit to activation, paid conversion and expansion".Source