Copilot Agents
Chat modes, instructions and prompt files for GitHub Copilot.
30 assets · curated & open-source
Accessibility Expert
Reviews code and designs against WCAG 2.1/2.2, covering semantics, keyboard and focus behavior, forms, media and dynamic SPA updates, and ships framework-specific examples plus a CI setup for automated checks.
Address Comments Agent
Works through pull request review comments one at a time: applies the minimal fix for each, pushes back on comments that don't make sense, adds test coverage, and commits with a descriptive message before moving to the next comment.
ADR Generator
Turns a decision discussion into a numbered Architecture Decision Record: gathers the missing context, assigns the next sequential ADR number, and writes a structured markdown file with consequences, alternatives and rejection reasons.
Blueprint Mode
Picks one of four fixed workflows (Main, Debug, Express, Loop) for a task, then executes it end to end with a numeric self-reflection rubric and a confidence threshold that decides whether to ask a clarifying question.
Code Tour Expert
Writes VS Code CodeTour .tour files: guided, step-by-step walkthroughs of a codebase with file references, directory steps and command links, aimed at making onboarding faster for new engineers.
Critical Thinking Mode
A Socratic questioning mode that refuses to write code or give direct answers. It repeatedly asks "why" about a plan, one question at a time, until the reasoning behind a decision is fully surfaced.
Debug Mode Agent
Runs a four-phase debugging workflow: reproduces the bug first, traces the root cause, applies a minimal fix, then verifies with tests instead of guessing at a patch.
Devil's Advocate
Argues against your idea on purpose, raising one objection at a time and refusing to propose fixes. Ends on request with a summary of how well the idea held up and which risks are still open.
Doublecheck Agent
Fact-checks AI-generated text before you act on it: extracts claims, searches for real sources, and runs an adversarial review for hallucination patterns, without rendering its own verdict.
GitHub Actions Expert
Reviews your GitHub Actions workflows against a security-first checklist, pinning commit SHAs instead of mutable tags, enforcing least-privilege permissions and OIDC over long-lived credentials, with built-in dependency, container and secret scanning.
Mentor Mode
Coaches instead of coding: it looks through the codebase to understand context, then challenges your assumptions with questions and the 5 Whys rather than editing files or handing you a solution.
Modernization Agent
A human-in-the-loop legacy modernization workflow that reads every service, repository and controller file in a codebase, documents each business feature in its own markdown file, then proposes a modern stack and a step-by-step migration plan.
One-Shot Feature Issue Planner
Turns a single feature request into a complete, issue-ready GitHub issue draft in one pass, no clarifying questions, inferring missing details from the codebase and labeling every assumption it makes explicitly.
Playwright Tester Agent
A Copilot custom agent that explores a website with the Playwright MCP server, then writes, runs and iterates on TypeScript Playwright tests until they pass.
PostgreSQL Database Administrator
Acts as a hands-on PostgreSQL DBA through the official VS Code PostgreSQL extension: connects to a real server, inspects schemas, runs and optimizes queries, and handles backups, instead of guessing from application code.
PRD Generator
Interviews you with 3-5 clarifying questions, reviews the codebase for constraints, then writes a full Product Requirements Document with user personas, functional requirements and testable user stories, and can turn each story into a GitHub issue on request.
Principal Software Engineer Agent
A senior engineering reviewer in the style of Martin Fowler: SOLID/DRY/YAGNI judgment calls, test-pyramid coverage, and mandatory GitHub issues for any technical debt it flags.
Project Architecture Planner
A cloud-agnostic architecture planner that runs a discovery interview, then delivers a tech stack evaluation, a phased scalability roadmap, a cost model and Mermaid/HTML/draw.io diagrams.
Prompt Engineer Agent
Treats every message as a prompt to improve rather than complete: analyzes it against a structured rubric, then rewrites it into a clearer, better-ordered system prompt.
SAST/SCA Security Analyzer
Runs static code analysis and dependency auditing in one pass, mapping every finding to a CWE ID, an OWASP 2025 category and policy frameworks like PCI-DSS, HIPAA and GDPR, with a prioritized remediation plan.
Security & OWASP Top 10 Instructions
Maps 55+ secure-coding anti-patterns to OWASP Top 10 2025 for Copilot, with detection regex and framework-specific fixes for Next.js, Angular, Express and Go, plus AI/LLM security guidance.
Specification Mode
Turns a piece of functionality into a formal, AI-readable specification file, with coded requirements (REQ/SEC/CON), Given-When-Then acceptance criteria and explicit dependencies, saved under /spec/ with a fixed naming convention.
Task Researcher
Research-only specialist that is not allowed to touch source code. It investigates a task using the codebase and external docs, then writes its findings to a dated file under .copilot-tracking/research, narrowing multiple approaches down to one recommended solution.
TDD Green Phase Agent
Second stage of the TDD workflow: writes the smallest amount of code that makes the current failing test pass, without touching the test itself.
TDD Red Phase Agent
First stage of a three-agent TDD workflow: writes one focused failing test at a time from GitHub issue requirements, and verifies it fails for the right reason before any implementation exists.
TDD Refactor Phase Agent
Third stage of the TDD workflow: cleans up duplication and design, and hardens security (input validation, secrets, OWASP) while keeping every test green.
Technical Debt Remediation Plan
Produces a markdown remediation plan for a piece of technical debt, scoring it on ease of fix, impact and risk of inaction, without making any code changes itself.
Terraform IaC Reviewer
Reviews and writes Terraform changes with a focus on state safety, least-privilege IAM, drift detection and a strict plan-before-apply discipline.
Thinking Beast Mode
A long, autonomous "keep going until it's solved" system prompt that pushes an agent through deep sequential thinking, recursive web research and rigorous self-testing before it stops working on a task.
Universal Janitor
Cleans up any codebase by deleting unused code, flattening unnecessary complexity and auditing dependencies, on the premise that removing code is the most effective refactor available.