The new strategic advantage isn’t buying AI — it’s controlling the stack it runs on. SOME1 forward-deploys a senior AI-engineering team to build custom software around your processes, with full control and governance over your models, harnesses and infrastructure. Shipped in weeks. Yours to govern.
This listing identifies a third-party MCP server. MCPBundles is not affiliated with, sponsored by, or endorsed by the service or its operator unless stated on this page.
This server may ask you to sign in when your AI client connects.
Last probed Nov 3, 2026 · mcp.some1.ai
12tools discovered
For AI agents — MCPBundles Directory API
For MCP client apps (human connect):
MCP connect URL: https://mcp.some1.ai/
Skill page: https://www.mcpbundles.com/skills/some1-mcp-5d2e453c9b
Explore capabilities (12)
read 6
write 1
other 5
Get
4
Other capabilities
8
List contracts you can publish against
read
list_contracts
START HERE when someone wants to sell a capability. A contract is the typed job spec — its input schema, output schema and the deterministic validator that decides whether a run gets paid. You publish AGAINST a contract; you do not invent one per offering, because buyers shop by contract and need interchangeable suppliers. Read the returned contracts to the human and work out which one their capability actually satisfies. If none fits, say so — that is a real answer, and the contract needs addin
Description may be truncated on the public preview.
List contracts you can publish against — START HERE when someone wants to sell a capability. A contract is the typed job spec — its input schema, output schema and the deterministic validator that decides whether a run gets paid. You publish AGAINST a contract; you do not invent one per offering, because buyers shop by contract and need interchangeable suppliers. Read the returned contracts to the human and work out which one their capability actually satisfies. If none fits, say so — that is a real answer, and the contract needs addin
Check what this account still needs before it can sell — Report where the human is in the publishing journey: whether their org is an approved publisher, whether Stripe Connect is attached (without it activation is REFUSED, because Some1 would have nowhere to send their 80%), what they have already listed, and each listing's status. Call this before and after any publishing step so you can tell them what is left rather than guessing. ECONOMICS: buyers prepay; Some1 keeps 20% of every settled execution and the publisher receives 80%, paid out through S
Store a credential Some1 sends when calling the capability — If the capability's endpoint needs a token, store it here and Some1 injects it as a header on every call. ASK THE HUMAN whether their endpoint requires auth before publishing — a listing whose endpoint 401s will fail conformance and they will not understand why. The value is WRITE-ONLY: it goes into Supabase Vault and neither you, nor this API, nor support can read it back, so tell them to keep their own copy. Never print the value back to the human in your reply, and never store it anywhere els
Set or change what a capability charges — Price a listing from MEASURED cost, never from a guess. Before calling this, help the human work out what ONE run of their capability actually costs them — model tokens AND any metered dependency (search, scraping, storage). Metered dependencies are the classic trap: they are billed separately from inference and are easy to leave out, which produces a listing that looks profitable and loses money on every sale. GUIDANCE: list at roughly 2.5x the measured floor. Some1 takes 20%, so the publisher
Check whether a listing has passed the fixture suite
Get your MCP into directories
A working endpoint is step one. Directory coverage is the coordinated launch across ChatGPT, Claude, Cursor, the MCP Registry, and community indexes.
The new strategic advantage isn’t buying AI — it’s controlling the stack it runs on. SOME1 forward-deploys a senior AI-engineering team to build custom software around your processes, with full control and governance over your models, harnesses and infrastructure. Shipped in weeks. Yours to govern.
How do I connect SOME1 to my AI agent?
Use the MCP endpoint listed on this page in your MCP client configuration. One-click install pills support Claude, Cursor, VS Code, and other hosts. Copy the remote MCP URL if your client needs a manual entry.
Operate this server?
Request removal if you do not want SOME1 listed here, or claim the listing if you want to manage it on MCPBundles.
Operate SOME1?
This server appears in the MCPBundles directory. Verify you operate it to take over the listing — name, description, logo, contact email, and skill content. We email a 6-digit code to a maintainer address your server publishes in /.well-known/security.txt or /.well-known/mcpbundles.json. Free, takes about a minute.
MCPBundles publishes this directory for MCP discovery. Except where we host or operate an offering, third-party services run under their own terms. Product and company names on this page are used in a descriptive, identifying way (including under nominative fair use where applicable); they remain the property of their owners. Nothing here grants you rights in those marks, and nothing here is an offer to sell a third party's services. Terms · Request removal
More in developer tools
Other MCP servers in this category from the directory index
— Conformance is the gate between "published" and "sellable": Some1 runs the contract's fixture suite against the real endpoint and an offering cannot go active until it passes. Use this to report progress and, when it fails, to tell the human WHICH fixtures failed and why — the report names each one. A failure is usually the endpoint being wrong, needing auth, or returning a shape the contract does not accept, and those are all fixable.
Search capabilities — Search the Some1 exchange for executable capabilities (active offerings). Filter by free-text query (case-insensitive match on contract name + display name; v1 filters in-process over the active list — Postgres FTS is deferred) and/or contract, sort by price/latency/success. Each result includes price_micro, `model` (the model powering it, e.g. "openai/gpt-5-nano" — null means the capability does no inference) with its model_provider, a 30-day reliability scorecard, and the publisher. The model
Describe capability — Full manifest for one active offering: `model` — the model that actually powers it (e.g. "openai/gpt-5-nano", "anthropic/claude-opus-4.7-fast") plus model_provider — then contract (name@major) with complete input/output JSON Schemas, price_micro, reliability scorecard, publisher org, and endpoint kind. READ THE MODEL FIRST: it is the single largest determinant of both what this capability costs and how deep its answers go — the spread between a small open model and a frontier one is roughly 1000
Get quote — Validate an input against the contract input schema and lock a price: returns a quote_id redeemable with execute_capability for 10 minutes. On invalid input the error text lists the exact schema violations so you can self-correct and retry. Works WITHOUT an API key (a quote reserves nothing), but an anonymous quote is a price check only — it comes back with redeemable=false, so mint a key with create_agent_account and re-quote before executing. Executions are charged from your prepaid Some1 cred
Execute capability — Execute a capability and pay for it from prepaid credits: either redeem a quote_id, or run directly with offering_id + input + max_price_micro (exactly one of the two forms). Reserves (holds) the price, starts the execution, then waits up to wait_seconds (default 30, max 60) for completion. If it finishes in the window you get the full execution detail (validated output as structured content on settled, plus the signed receipt_id); otherwise { execution_id, status } to poll via get_execution. Re
Get execution — Fetch the current state of an execution: status, validated output (when settled), error (when failed/refunded), attempts, timings, and the receipt_id of the signed receipt proving what ran and what was charged from your prepaid credits. Executions are charged from your prepaid Some1 credits (price_micro = micro-USD, 1,000,000 = $1), and every settled result carries a verifiable cryptographically signed receipt (receipt_id).
Cancel execution — Best-effort cancel: a still-reserved execution is refunded (the credit hold is released back to your prepaid balance); anything already running or finished is untouched and its current status is returned. Settled executions keep their signed receipt (receipt_id). Executions are charged from your prepaid Some1 credits (price_micro = micro-USD, 1,000,000 = $1), and every settled result carries a verifiable cryptographically signed receipt (receipt_id).
Create agent account — Mint your own Some1 agent account in ONE call — no human, no signup form, no API key needed to call this. You get back: agent_id (your org on the exchange), api_key (shown ONCE — store it and send it as `Authorization: Bearer <api_key>` on every later request), a trial credit of 500000 micro-USD ($0.50, enough for roughly a handful of executions), your wallet balance, and the mcp_url/api_url to keep using. TRIAL CREDITS ARE LIMITED AND NOT REFILLED: when they run out, execute_capability returns
MCPBundles probed 12 tools on the live server. The tool list on this page reflects what was discovered at the last refresh — connect your client to see the full set available to your session.
What authentication does SOME1 require?
SOME1 may require signing in to the provider before tools can run. Connect through MCPBundles or your MCP client and complete any provider login when prompted.
Don't want this listed?
Request removal from the MCPBundles directory. We stop publishing this listing on mcpbundles.com.
MCPBundles is an independent platform built on the open Model Context Protocol standard. MCPBundles is not affiliated with, sponsored by, or endorsed by Anthropic PBC.