openclaw-workspace

Summary

This workspace provides an AI-driven environment for analyzing reporting data, conducting BigQuery analyses, and managing todos. The focus is on Metabase reports, deep analysis of user journeys and issues for a financial app using BigQuery, and organizing operational or agent workflows. Documentation, agent memory, and project reference materials are included.

Included Assets

• AGENTS.md: Agent operational instructions and memory management guides.
• MEMORY.md: Workspace long-term memory and essential business definitions.
• SOUL.md: Agent guidelines, personality, and behavioral boundaries.
• TOOLS.md: Local environment, device, and tool configuration notes.
• docs/
  • emma-analysis/: Full stack of markdown reports on user lifecycle, root cause, triggers, and user journeys around "919" carrier verification error.
    • README.md, 919_root_cause_analysis.md, 919_trigger_monthly.md, 919_user_journey.md
  • DEFINITIONS.md: Definitions for audience dashboard metrics and 919 error criteria.
  • audience_dashboard.html: Static dashboard for retention cohort analysis.
• invoice-prototype/
  • README.md: Details about the AI-powered finance insight app prototype.
  • package.json: React/Firebase app package and scripts.

How to Use

1. Agent Operation:

   • Begin each session by reviewing SOUL.md for agent persona and workspace rules.
   • For context and continuity, consult AGENTS.md for memory instructions, and use MEMORY.md in main sessions to keep track of significant events and curated knowledge.

2. Data Analysis:

   • BigQuery SQL logic and metric definitions are in docs/DEFINITIONS.md and docs/emma-analysis/ subfolder.
   • Explore root cause and lifecycle analyses of app user retention and "919" verification issues via the markdown files in docs/emma-analysis/.
   • For dashboard usage, reference docs/audience_dashboard.html (static HTML).

3. Prototype Application:

   • The "invoice-prototype" directory contains a React app for financial insights.
   • Install dependencies and run scripts as specified in invoice-prototype/package.json to develop or preview the prototype app.

4. Customization:

   • Record local hardware, SSH, or environment specifics in TOOLS.md.
   • Update memory and operational notes as agent sessions progress.

Notes

• This workspace is designed for agents and users who need reliable, reproducible setups for data-driven analysis and reporting, especially with rapidly changing operational or technical context.
• The internal memory system uses both daily logs and curated summaries for context retention.
• All analysis and process documentation is provided in Markdown or HTML for ease of review and replication.
• The React prototype app uses Firebase for real-time data and can be run locally or deployed.
• Most documentation and analysis is targeted at a Mandarin-speaking audience, with a focus on Taiwan e-invoicing and app lifecycle issues. Some files include sample SQL queries and metric tables for advanced analytics usage.
• No backend orchestration or external automation tooling is bundled; external connections (such as to Metabase or BigQuery) are expected to be managed via your own credentials and infrastructure.