Terms of Service

1. Acceptance of Autonomous Runtime Governance

GnamiSI provides an autonomous AI agent environment capable of planning workflows, modifying workspace files, and issuing terminal commands. You acknowledge that autonomous agents operate based on statistical probability and require human oversight. You agree to review all Human-in-the-Loop (HITL) gates before authorizing destructive system actions.

2. Model Access

Local agent runs start the Codex CLI for supported OpenAI models or the Grok CLI for supported xAI models on your computer. Each CLI uses the sign-in configured in that program. GnamiSI does not receive, store, or proxy those credentials.

GnamiSI does not collect, store, or use provider API keys. Cloud runs use the Cloudflare Worker model list (Workers AI models, and Gnami with separate token billing).

You must follow the terms of the model vendors whose CLIs or cloud models you use.

3. PayPal Exclusive Subscription & Quotas

The GnamiSI Pro tier ($29/month, 12,500 cloud credits per month) is managed exclusively via PayPal. Subscriptions automatically renew monthly until canceled. Compute quotas and cloud credits are provisioned immediately upon PayPal transaction confirmation. Each month the balance resets to the plan allowance: 12,500 while a Pro subscription is active, including when Lifetime is also owned, and 2,500 for Lifetime once Pro has ended. Unused credits do not roll over. A canceled, suspended, or expired Pro subscription does not reset. Usage through your own model subscriptions (Codex, Grok, etc.) never consumes credits. A completed cloud run costs a flat 20 credits (10 for planning and 10 for execution), charged after the run succeeds. If the balance is lower than 20, it is reduced to zero rather than below zero. Failed runs, model errors, and timeouts cost 0 credits. Starting a cloud run requires an active Pro or Lifetime subscription and a balance of at least 10 credits. Runtime is capped at 120 seconds. Cloud runs are limited to 20 per user per UTC day. Failed runs caused by model or provider errors do not count toward the 20-run daily limit; timeouts do. Separately, you can start at most 40 cloud runs per UTC day, and at most 5 per minute. Brief chit-chat costs 2 credits after a successful reply, and 0 credits if that model request fails. Chit-chat does not count toward the 20-run daily limit.

Gnami inference uses a separate prepaid wallet and does not deduct cloud credits. The wallet is stored in microdollars, where 1 USD equals 1,000,000 microdollars. A completed $10.00 PayPal purchase adds 10,000,000 microdollars. Before the model call, the Worker reserves the maximum charge for that request: 32,768 input tokens at the configured input rate, plus the requested output-token limit at the configured output rate. After a successful response it charges the actual token counts. Input tokens and output tokens are each multiplied by their configured USD-per-million rate and rounded up to a whole microdollar. If the request fails, the reservation is refunded and nothing is charged. Those rates are Worker environment settings. This section does not set a fixed public token price.

Lifetime Access buyers request a refund through the contact form at gnamiai.com/contact within 30 days of purchase. Refunds are paid back to the original PayPal payment.

4. Cloudflare Edge Rate Limiting & Zero Abuse

To maintain system uptime for all developers, Cloudflare Workers enforce edge rate limits on login and API endpoints. Automated scraping, credential stuffing, denial-of-service attempts, or bypassing the rate limit cooldown banner will result in permanent edge IP firewall blacklisting.

5. Limitation of Liability

GnamiAI (the company that operates GnamiSI) and its contributors provide the runtime on an “AS IS” basis. In no event shall GnamiAI be liable for code regressions, data loss, or server downtime resulting from agent execution. Users are advised to rely on the Time-Machine Git wrapper to preserve micro-snapshots.