GitHub Copilot

Public samples and documented tools that use GitHub Copilot in agent, CLI, and editor tooling.

GitHubCopilotSDK

GitHub Copilot SDK - same session API in C#, TS, Python, and Go

GitHubCopilotSDK is a multi-language learning repo that runs the same Copilot SDK session flow in C# (.NET 10), TypeScript, Python, and Go - all on SDK 1.0.14. Each CLI creates a session (model auto, approve-all permissions), sends a simple prompt, and prints the assistant reply. The Go sample highlights CLI-path discovery because that SDK does not bundle a runtime; Python’s runner bootstraps a venv and avoids the Windows Store stub. Aimed at engineers comparing Copilot SDK ergonomics across languages for agent or CLI tooling.

NABS Retros

NABS Retros - a VS Code extension that keeps agile retrospectives in the repository

A VS Code extension I designed and built in TypeScript that runs agile retrospectives inside the team’s own Git repository. It adds a Retros view, a Home page and a webview Retro Dashboard that steps each retro through six stages mapped to the retrospective maturity curve (Prepare and Gather are fully guided so far), with retros, feedback and objectives stored as plain Markdown that is committed and reviewed like the rest of the code. Objectives carry acceptance criteria with automatic measures from Git and GitHub data; GitHub Copilot is optional and only suggests summaries and assessments, including through the @retro chat participant, while people record the verdict. Action items become GitHub issues that the next retro checks, and finished retros can be archived to a GitHub Discussion. Distributed as a VSIX rather than through the Marketplace. The source is private; the public user guide (checked against v0.11.0) shows how it works.

efcore-sargable-config

EF Core + SQL Server - a Copilot agent skill for non-sargable queries, backed by plan-asserting tests

efcore-sargable-config is a GitHub Copilot agent skill for finding why an indexed EF Core / SQL Server lookup scans. Stable rules ESC001-ESC012 cover varchar/nvarchar and MaxLength mapping mismatches, blanket ConfigureConventions defaults, and non-sargable predicates such as functions or arithmetic on a column, implicit conversions, leading wildcards, OR across columns and ISNULL/COALESCE; every finding uses a fixed format, is labelled confirmed, candidate, other-cause or intentional exception, and a scan is only claimed when an actual execution plan shows it. A .NET 10 / EF Core test harness backs the guidance: seeded bad, good and clean C# cases for each rule with a scoring manifest, xUnit unit tests for effective mappings, generated SQL and a model-vs-schema guardrail, and Testcontainers integration tests that run every case against SQL Server 2022 and assert the actual plans (CONVERT_IMPLICIT, Seek vs Scan, logical reads, collation). The skill itself stays guidance: it does not change mappings automatically or connect to your database.