Copilot-Stuff

LLM-as-judge scoring skill. Takes a transcript + sandbox state, applies the rubric, and produces a structured score report in docs/testingResults/.

Iterative prompt improvement loop (Karpathy autoresearch pattern). One iteration per invocation — run → score → hypothesize → apply → re-test → commit if improved, revert if regressed. Human-supervised; designed for single-turn skills.

Test distributable Copilot artifacts (agents, prompts, instructions, skills) against the testing_sandbox workspace. Use when you want to validate real behavior, collect actionable feedback, and publish dated results in docs/testingResults/.

Synchronize AGENTS.md content into all detected AI coding agent configurations. AGENTS.md is the universal source of truth — read natively by GitHub Copilot (Aug 2025+), OpenAI Codex, Cursor, Gemini CLI, and VS Code. Falls back to .github/copilot-instructions.md for repos that have not migrated yet. Use when asked to mirror AGENTS.md or .github/copilot-instructions.md into Claude, Cursor, Gemini, Windsurf, and related tooling.

Synchronize GitHub Copilot instructions, custom agents, and skills into detected AI coding agent configurations in this repository. Use when asked to mirror .github/copilot-instructions.md, .github/instructions, .github/agents, or .github/skills into Claude, Codex, Cursor, Gemini, Windsurf, and related tooling.

Synchronize AGENTS.md content into all detected AI coding agent configurations. AGENTS.md is the universal source of truth — read natively by GitHub Copilot (Aug 2025+), OpenAI Codex, Cursor, Gemini CLI, and VS Code. Falls back to .github/copilot-instructions.md for repos that have not migrated yet. Use when asked to mirror AGENTS.md or .github/copilot-instructions.md into Claude, Cursor, Gemini, Windsurf, and related tooling.

Automate Dependabot PR merging safely, align branch protection to current CI jobs, and remediate stored-XSS output sinks. Use when asked to set up Dependabot auto-merge, harden branch protection for CI, or fix output-encoding vulnerabilities in URL composition, templated script output (e.g. JSON-LD), or RSS/XML interpolation. Works with any GitHub-hosted repository.