issues Search Results · language:Dune language:Python language:HTML language:Java language:Python language:Python
Filter by
42.2M results
Document setup, architecture, design decisions, SQL safety model, limitations, extension points, and sample demo
questions so the system is easy to review and discuss.
Add planner, schema, SQL, and summarizer agents. Use Gemini for live runs and a fake provider for deterministic tests.
Route all tool use through the adapter and MCP router.
Implement async routing from normalized tool calls to the correct MCP server over stdio. Include timeouts, structured
errors, and tests for dispatch behavior.
Define provider-agnostic tool spec, call, and result types. Add Gemini and OpenAI-style adapters so provider-native tool
schemas can map into the same internal protocol.
Expose read-only SQL validation and execution through an MCP stdio server. The server should use SQLite read-only mode,
enforce row limits, and return structured results/errors.
Implement a SQL validation layer for generated queries. It should parse SQL, reject writes and multiple statements,
enforce read-only SELECT behavior, and validate referenced tables.
Expose database metadata through a dedicated MCP stdio server. It should provide table listing, table descriptions,
database summary, and source lineage without executing arbitrary SQL.
Add basic command line entry points for building the database and inspecting table counts/schema. This gives reviewers a
simple way to verify the local data foundation.
Add deterministic seed data and a builder script so the local FieldOps database can be rebuilt from scratch without
network access. The builder should leave room for public data refreshes later.

Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.