Privacy and Data

1. Local project operations

Your project files, .engineforgeignore processing, local project index, Code Graph source data, and Unity/Godot editor bridges operate on the development machine. The bridge connects EngineForge to the local game editor.

2. Remote model inference

When you submit a Crew task, EngineForge sends the selected model, system instructions, conversation messages, tool definitions, and selected project or engine context through the EngineForge gateway. The gateway authenticates the account and routes the request to the configured model provider, which today is Anthropic, OpenAI or DeepSeek depending on the model you select. DeepSeek is a Chinese company, so a request routed to a DeepSeek model is processed outside the EU, the UK and the US. Request and usage metadata can include user, conversation and session IDs, model, token usage, tool and bridge labels, tool errors, EngineForge/plugin/engine versions, and operating-system platform.

3. Automatic service telemetry

• Bridge command telemetry: engine, command labels, success/failure, bounded error text, and optional session/conversation IDs.
• Search quality telemetry: the search query, result and candidate counts, score statistics, redirect state, latency, engine, and optional session/conversation IDs.
• Index health telemetry: engine type and aggregate class, function, edge, embedding, coverage, parse-error, and duration statistics.

4. User-submitted reports and feedback

• Session feedback sends like/dislike, optional text, session/conversation IDs, prompt count, model, engine, and EngineForge version when you submit it.
• A diagnostic report can include the capped conversation, log buffer, tool timeline, agent state, memory use, versions, identifiers, engine/model data, and your note. Diagnostic blobs are uploaded to EngineForge-managed Supabase Storage; report metadata is stored separately.

Website analytics

The website loads Vercel Analytics globally in production for page-view and route-path measurement. The inspected integration does not configure custom analytics events or a before-send transformation.

Authentication and account data

Website and platform service requests use account authentication backed by Supabase. GitHub and Google sign-in use their OAuth flows and return to the EngineForge authentication callback. EngineForge reads the resulting user identifier, email, session, and available profile metadata such as display name and avatar for account UI, access checks, and authenticated service requests.

Service storage

Operational data described above is stored in EngineForge service tables, storage, and structured logs where the implementation records it. OAuth providers and Supabase are separate service destinations in the authentication path.

Policy details not yet published

The inspected implementation does not establish public commitments for retention periods, provider training use, deletion timing, encryption guarantees, data residency, or regional processing. These points require approved policy text. For privacy questions, contact support@engineforge.ai.

Related documentation