Onboarding prompt

Help me onboard to Altertable and answer a first question with my data. 1. Initialize the Altertable connection before using other Altertable tools. Confirm the organization and environment. If access is missing, follow https://altertable.ai/docs/agents/setup.md and the matching client guide. Preserve existing configuration and use the normal browser authorization flow. 2. Ask what I want to answer or build. Ask one question at a time. If you have project access, inspect the relevant project files to discover the application language, databases, analytics SDKs, and OpenTelemetry configuration. Do not read secret files or print credentials. Ask only for information you cannot discover. 3. Learn where the relevant data lives, whether it needs to stay there, and how fresh the answer needs to be. Summarize the goal and stack before proposing a route. 4. Read https://altertable.ai/docs/lakehouse/ingest-or-connect.md. Recommend the shortest supported route: connect an external catalog to query data in place, ingest records or files, instrument Product Analytics, export OpenTelemetry logs, traces, or metrics, or combine routes. Explain what will be copied and what will stay in its source. Use current docs and available tools rather than inventing connector capabilities. 5. For terminal access, file ingestion, or scripts, check whether the Altertable CLI is installed. Follow https://altertable.ai/docs/lakehouse/clients/cli.md to install it when needed. Ask me to complete browser login, inspect the active profile, and verify the target environment. Use --agent for structured output. MCP and CLI have separate authentication; verify both when using both. 6. Explain the proposed setup and get my approval before creating persistent resources, changing application code, or sending data. Do not overwrite data, deploy, or publish anything as part of onboarding. Keep credentials out of the conversation and source files. 7. Connect one source or ingest a small, relevant dataset. Inspect actual catalog and table schemas. Run a bounded query and verify the returned rows, fields, and timestamps. An empty result is not proof that ingestion worked. If data is missing, diagnose the setup before continuing. 8. Answer my original question using the verified data. State the source, filters, time range, and any limitations. If the question needs unavailable data, explain the gap instead of guessing. 9. Report what you configured, what you verified, what I still need to do, and one useful next step. If a restart is required, give me a short handoff to paste into the next session. If you lack terminal or project access, guide me through the equivalent app steps or provide commands for my shell.