SaSameFor people and AI systems
Search knowledgeStart with SaSame
publisheddocumentationhowto

How do I start using SaSame?

One introduction and installation hub for humans, developers and AI agents using SaSame.

Collection
documentation
Updated
2026-08-24
min read
5
Verified summary

Connect an AI, choose what it may use, let it work through MCP, and review what happened. Start with the free path: audit a public MCP, connect an AI client, start a Marketing Mission, or run the local CLI.

GO

Start and install

10

Audit my MCP — free

Run the live point-in-time audit without an account or payment.

Open

Connect an AI client

Use the client-specific ChatGPT, Claude, Claude Code or generic MCP guide.

Open

Start through the Factory MCP

Connect an AI client, call start_here, then use the live Factory tools returned by tools/list.

Open

Open public MCP endpoint

The keyless Streamable HTTP endpoint used by AI clients.

Open

Read the CLI guide

Run the same bounded readiness method locally.

Open
ChatGPT

Connect in ChatGPT

Official guide ↗

Create a custom MCP app/connector in ChatGPT developer mode and paste the SaSame remote endpoint. Current plan and workspace permissions are controlled by OpenAI.

Remote MCP support and developer-mode permissions vary by ChatGPT plan and workspace role.
Endpoint
https://live-vps.sasame.online/public-mcp
Configuration
https://live-vps.sasame.online/public-mcp
  1. 1

    Open ChatGPT Settings or Workspace Settings and locate Apps / developer mode.

  2. 2

    Create a custom MCP app or connector.

  3. 3

    Paste the copied SaSame endpoint and scan or refresh tools.

  4. 4

    Enable the app in a chat and call start_here.

Verify

Start a chat with the app enabled and ask it to call start_here. Confirm that the returned machine map names the SaSame public surface.

Claude

Connect in Claude

Official guide ↗

Add SaSame as a custom connector using the remote MCP URL in Claude or Claude Desktop.

Custom remote connectors are available on supported Claude plans; organization owners may need to add them first.
Endpoint
https://live-vps.sasame.online/public-mcp
Configuration
https://live-vps.sasame.online/public-mcp
  1. 1

    Open Settings > Connectors.

  2. 2

    Choose Add custom connector.

  3. 3

    Paste the copied SaSame endpoint and add the connector.

  4. 4

    Enable it from Search and tools, then call start_here.

Verify

Enable the connector from Search and tools, then ask Claude to call start_here and report the available shortest path.

Claude Code

Add with one command

Register the remote Streamable HTTP endpoint from the Claude Code CLI.

Endpoint
https://live-vps.sasame.online/public-mcp
Command
claude mcp add --transport http sasame https://live-vps.sasame.online/public-mcp
  1. 1

    Copy and run the command in a terminal.

  2. 2

    Confirm that sasame appears in `claude mcp list`.

  3. 3

    Call start_here from a Claude Code session.

Verify

Run `claude mcp list`, start Claude Code, and ask it to call start_here.

Generic MCP client

Copy JSON configuration

Use this configuration in clients that accept a Streamable HTTP MCP server definition.

Endpoint
https://live-vps.sasame.online/public-mcp
Configuration
{
  "mcpServers": {
    "sasame": {
      "type": "http",
      "url": "https://live-vps.sasame.online/public-mcp"
    }
  }
}
  1. 1

    Copy the JSON configuration.

  2. 2

    Paste it into the client's MCP configuration surface.

  3. 3

    Restart or refresh the client if required.

  4. 4

    Call start_here.

Verify

Refresh the client's tool list and call start_here. tools/list is authoritative for the current surface.

Terminal / npm

Run the readiness CLI

Audit a public MCP locally with no account or SaSame credential.

Command
npx -y mcp-readiness@latest gold-rush start <your-mcp-url> --json
  1. 1

    Replace <your-mcp-url> with a public MCP endpoint.

  2. 2

    Run the command locally.

  3. 3

    Save the JSON result as the baseline for later comparison.

Verify

Confirm the command returns JSON containing the bounded readiness grade and criterion results.

01

Audit an MCP without installing anything

Open the free audit, paste a public MCP endpoint and receive a point-in-time readiness result. No account, payment, claim or subscription is created by the audit.

02

Connect SaSame to an AI client

Use the copy controls below for ChatGPT, Claude, Claude Code or another Streamable HTTP MCP client. The public endpoint is keyless. Client availability and workspace permissions still depend on the client provider.

03

Start a Marketing Mission

Use the authenticated account MCP when you want SaSame inside your AI client. The Marketing Mission tools help you define the audience, channel and offer, check public evidence, and draft a human-reviewed message or campaign asset. They do not send messages, post to external sites, scrape private contacts, or run bulk outreach.

04

Run the open-source CLI

Use mcp-readiness locally when you want a reproducible readiness grade without connecting an AI application or publishing a result.

05

Use the free beta path

There are three simple paths, depending on what you need first.

  1. No account: connect the public MCP at https://live-vps.sasame.online/public-mcp, call start_here, then call audit_mcp or factory_start.
  2. Site knowledge: connect the knowledge MCP at https://srl-sasame.com/api/mcp and call knowledge_search or knowledge_get.
  3. Account tools: sign in at Account Control, connect the authenticated Client MCP at https://live-sasame.sasame.online/mcp, then use capability_set, usage_reset, marketing_mission_start, marketing_mission_evidence, marketing_mission_draft, factory_dashboard and factory_export.

During Capability Control Beta, new paid Factory checkout is disabled. factory_membership_checkout is present for compatibility and existing-subscriber continuity, but new users do not need it and should not start there.

06

Enter the Factory through MCP

Connect an AI client to the public MCP and call start_here to see what is available. Use factory_start for the free Factory lifecycle entry that matches your state: idea, blueprint, unfinished build, finished build, live MCP, underperforming MCP or portfolio monitoring. If you need account-scoped dashboard/export tools later, sign in at Account Control and connect the authenticated Client MCP.

EX

Examples

5
  1. 01

    After connecting the public MCP, call start_here and then audit_mcp with a public endpoint URL.

  2. 02

    After connecting the authenticated account MCP, call marketing_mission_start, then marketing_mission_evidence, then marketing_mission_draft.

  3. 03

    After running the CLI, keep the JSON output as a local reproducible baseline.

  4. 04

    After signing in, keep all beta capabilities enabled by default or call capability_set to hide anything your AI client should not see.

  5. 05

    If the beta bucket reaches quota, reset it from Account Control or call usage_reset; lifetime history is preserved.

RF

References

6
NX

Continue exploring

6
documentation01

Getting Started

Identify your MCP state, select the matching Factory entry point, then collect a reproducible baseline before changing anything.

documentation02

mcp-readiness CLI

mcp-readiness is SaSame’s open-source command-line tool: run it with npx against any public MCP endpoint to get the same bounded readiness grade the Observatory publishes, with no account or key required.

products03

Products

SaSame offers one MCP Factory, and CP4 now keeps product decisions tied to independent demand evidence: technical readiness, simulations, crawler traffic or internal confidence cannot create SCALE or permanent SKU status.

pricing04

Pricing

Capability Control Beta is the current new-user path: sign in, connect an AI client, choose capabilities, use them — no payment required for new users during beta. New paid Factory membership checkout is disabled; Free Listing remains available. Existing subscribers retain all entitlements, renewals and webhooks. CP4 keeps North-Star accounting strict: only independent LIVE payment plus exact verified fulfillment can count; checkout starts, tests, crawler signals and synthetic replays never count.

documentation05

Agent Execution Router

Call resolve_and_run on the same public MCP endpoint with a goal, environment and constraints; it returns a bounded route and an honest execution state — only "verified" means something actually ran.

documentation06

Connect an AI Client

Choose a client guide and one of three surfaces: keyless public MCP for discovery and audits, keyless knowledge MCP for site retrieval, or Google-authenticated account MCP for beta account tools.