Skip to content
CAI
Software that uses CAICheck a score

SebastienDegodez/copilot-instructions

48.9

Weak · 9 November 2026

2.2k

lines of production code

TypeScript

with Python, C#

7

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

New C\# Clean Architecture development plugin with testing and architecture skills

Added a new Copilot plugin for C\# developers that provides guidance on Clean Architecture, Domain-Driven Design, and CQRS patterns. The plugin includes two main skills: 'clean-architecture-dotnet' for setting up layered .NET projects without MediatR (using convention-based DI and ArchUnit validation) and 'application-layer-testing' for implementing sociable unit tests with mutation testing (Stryker). It also introduces placeholder directories for global instructions and plugins to support future extensibility.

agents, instructions, plugins · high confidence

New Superpowers Whetstone plugin for outside-in TDD and mutation testing

The new \superpowers-whetstone\ plugin introduces a structured testing workflow for behavior-first development. It includes three core skills: \gherkin-gate\ to capture and approve business scenarios before implementation, \outside-in-tdd\ to drive design from failing tests using real domain objects and mocked external boundaries, and \mutation-testing\ to validate test effectiveness using Stryker.NET. The plugin also provides CQRS patterns, sociable testing templates for command and query handlers, and a detailed testing strategy guide.

plugins/superpowers-whetstone · high confidence

New autoresearch skills for Clean Architecture, Mutation Testing, and Outside-In TDD

Added three new autonomous skill optimization loops under .autoresearch: 'clean-architecture-dotnet' (guiding .NET layering, CQRS, and DDD patterns), 'mutation-testing' (validating test quality via Stryker.NET), and 'outside-in-tdd-v2' (refining test-driven development workflows). Each skill includes a baseline SKILL.md, a changelog documenting iterative improvements, a results dashboard (HTML/JS) for monitoring experiment progress, and structured result files (JSON/TSV) tracking pass rates and evaluation metrics.

.autoresearch · high confidence

Behavioural changes

2 commits (0 fixes) modifying skills

A change to existing behaviour in skills — 2 commits, 20 files: “feat: Update SKILL.md (\#25)”; “Refactor/copilot instructions (\#16)” (refs \#25, \#16).

skills · medium confidence · unverified

3 commits (0 fixes) modifying plugins/minimal-context-tools

A change to existing behaviour in plugins/minimal-context-tools — 3 commits, 24 files: “feat: Update snip.json to change hook commands”; “feat(minimal-context-tools): ship snip pre-tool-use hook (\#35)”; “feat: add minimal-context-tools plugin with snip hooks and dotnet filters” (refs \#35).

plugins/minimal-context-tools · medium confidence · unverified

Clean Architecture .NET skill updated to version 2.0

The C\# Clean Architecture development skill has been updated to version 2.0, introducing a revised project structure and testing strategy. The previous 'application-layer-testing' skill and its associated templates (CommandHandlerTestTemplate, QueryHandlerTestTemplate) and reference guides (CQRS patterns, mutation testing, sociable testing) have been removed. They are replaced by a new 'clean-architecture-dotnet' skill that enforces a strict four-layer architecture (Domain, Application, Infrastructure, API) with marker interfaces for assembly identification. The new skill promotes a MediatR-free CQRS implementation using explicit \ICommandHandler\ and \IQueryHandler\ interfaces, convention-based dependency injection, and a test-first workflow validated by mutation testing (Stryker.NET). Also in this area: New C\# Clean Architecture development plugin with testing and architecture skills.

plugins/csharp-clean-architecture-development · high confidence

Evaluator now preserves benchmark history and improves discovery reliability

The evaluator tool now maintains a history of past benchmark results for each entry, allowing users to track performance trends over time rather than seeing only the latest snapshot. Discovery of evaluation entries has become more deterministic and efficient by skipping assets or instructions that lack corresponding test scenarios, and by explicitly excluding GitHub workflow files from the evaluation scope. Additionally, the tool now handles invalid commit SHAs during incremental discovery by throwing clear errors or falling back to re-evaluating the entry, and benchmark reporting provides feedback on whether the summary was actually updated. Also in this area: Evaluator adds tool-calling support and identifies itself to LLM providers; Introduce automated evaluation pipeline for skills, plugins, and instructions.

tools/evaluator · high confidence

Improved skill deduplication and unified install bars with VS Code support

The website now prioritizes root skills over plugin-bundled duplicates and deduplicates slugs, ensuring the most relevant skill page is displayed. Install instructions on agents, instructions, plugins, and skills pages have been replaced with unified tabs: agents and instructions now offer APM and VS Code (including Insiders) installation options, plugins show Copilot CLI and APM commands, and skills display an APM command with a direct link to view the file on GitHub. Additionally, the home page has been simplified by removing the Browse Skills and View Instructions buttons, and the plugins page now reads README files from the correct relative path. Also in this area: Remove slash-command label from benchmark source types; New Benchmarks dashboard and inline evaluation panels; Fix base path in Astro configuration to include trailing slash.

website/src · high confidence

Fixes

Fix base path in Astro configuration to include trailing slash

The \base\ configuration in \website/astro.config.mjs\ has been corrected to include a trailing slash (\/copilot-instructions/\) in production builds. This ensures that relative asset paths and navigation links resolve correctly when the site is deployed to a subdirectory, preventing broken links and missing resources for users.

website · high confidence

Test coverage

Added tests for evaluator tool discovery and benchmark reporting

Added comprehensive test coverage for the evaluator tool's core logic and CLI commands. Tests verify that the benchmark summary merge function correctly handles appending history, creating new entries, and preserving existing data without mutation. Tests for the discover command and discoverer module confirm that the tool correctly identifies evaluable assets, treats evaluation workflow changes as infrastructure changes, and handles scenarios where previous commit SHAs cannot be resolved. Additionally, tests for the report command ensure that summary.json is updated only when fresh results are provided and that new entries are created when necessary. Also in this area: Added tests for LLM client tool completion methods; Added unit tests for the evaluator tool's discovery, judging, and execution logic.

tools · high confidence

Dependencies

Added automated evaluation pipeline tool

Introduced a new \@copilot-instructions/evaluator\ tool in the \tools/evaluator\ directory, providing an automated evaluation pipeline for Copilot skills, plugins, and instructions. This addition includes the project manifest (\package.json\) and lockfile (\package-lock.json\), establishing dependencies such as \openai\, \commander\, \zod\, and \pino\, along with development tools like \vitest\ and \eslint\ to support the new capability. Also in this area: Add @tailwindcss/typography dependency; Initial website dependency setup with Astro, Tailwind, and Marked.

(dependencies) · high confidence

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

How this codebase got here

Score

  • CAI 51 → 49 (-1.9)
  • Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

Lenses

  • Code Health 68 → 68 (+0.0)
  • Architecture 94 → 94 (+0.0)
  • Maturity 72 → 73 (+1.6)
  • Readiness 50 → 46 (-4.7)
  • Security 70 → 72 (+2.6)
  • Accessibility 41 → 39 (-1.8)
  • Performance 60 → 60 (+0.0)

New (20)

  • Banned license: @img/sharp-libvips-darwin-arm64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-darwin-x64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-arm (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-arm64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-ppc64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-riscv64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-s390x (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linux-x64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linuxmusl-arm64 (website/package-lock.json)
  • Banned license: @img/sharp-libvips-linuxmusl-x64 (website/package-lock.json)
  • Banned license: @img/sharp-wasm32 (website/package-lock.json)
  • Banned license: @img/sharp-win32-arm64 (website/package-lock.json)
  • Banned license: @img/sharp-win32-ia32 (website/package-lock.json)
  • Banned license: @img/sharp-win32-x64 (website/package-lock.json)
  • High CVE: [GHSA redacted] (tools/evaluator/package-lock.json)
  • Medium CVE: [GHSA redacted] (tools/evaluator/package-lock.json)
  • Medium CVE: [GHSA redacted] (tools/evaluator/package-lock.json)
  • Medium CVE: [GHSA redacted] (tools/evaluator/package-lock.json)
  • Orphaned file (tools/evaluator/src/reporters/markdown-reporter.ts)
  • Orphaned files with no living knowledge

Changes since last survey

  • 1 commits — 1 feature/other, 0 fixes

By area

  • tests/plugins — 1 commit

Notable commits

  • change: feat(plugins): add dual-client LSP plugins for Copilot CLI and Claude Code (#37)

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

Survey your own repository

SebastienDegodez/copilot-instructions was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.

About this page

  • The score is its most recent published measurement, taken on 9 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
  • Measured at commit fa2d13ae8af5037c2b658ea7459ea275975ec7dd — the exact code this score is about.
  • Scored under rubric-2026.11.2 — the same rubric and the same method as every other entry in this index.
  • Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.3+preprod-bb0cb800cd0f.