dongbumlee/sdlc-harness
SDLC Harness — an agent-driven multi-agent system that orchestrates your entire software development lifecycle with harness-quality QA, 19 specialized agents, and 16 domain skills.
Create Architecture Decision Records following SDLC standards and application templates. Use when documenting design decisions, architecture choices, or technology selections. Triggers on ADR, architecture decision, design proposal, or design document requests. Combines awesome-copilot ADR skill with SDLC-specific ADR template from .design/ADR-TEMPLATE.md.
Review code for architecture and design consistency following SDLC layering rules and application project patterns. Use when reviewing PRs, checking dependency boundaries, or validating pattern reuse. Triggers on architecture review, layering check, design review, or pattern compliance requests.
Create Azure infrastructure with Bicep using Azure Verified Modules (AVM), configure azd orchestration, and manage deployment lifecycle. Use when writing Bicep templates, configuring azure.yaml, setting up Container Apps, or preparing deployments. Triggers on Bicep, AVM, azd, infrastructure, or deployment requests.
Implement Azure Blob Storage and Queue operations using the approved Storage library. Use when uploading, downloading, listing, or managing blobs and queues. Triggers on blob, storage, file upload, queue, or Azure Storage requests. Never use raw azure-storage-blob or azure-storage-queue — always use the approved Storage library.
Instructs the Harness agent how to run E2E canary tests against SDLC phase agents. Canary mode validates that the harness orchestration is working correctly by feeding known prompts to phase agents and checking outputs against expected criteria.
Review and enforce code quality standards for Python, TypeScript, React, Java, C#, Go, and Rust following SDLC quality instruction files. Use when performing code quality reviews, cleaning up code, or enforcing naming and documentation standards. Triggers on code quality, code review, naming, docstring, dead code, or comment cleanup requests.
Implement Azure Cosmos DB data access using the approved Cosmos DB library with Repository Pattern. Use when creating entities, repositories, or any Cosmos DB CRUD operations. Triggers on Cosmos DB, database, entity, repository, data model, or data access requests. Never use raw azure-cosmos SDK — always use the approved Cosmos DB library.
Generate and read the project manifest that records which templates and patterns were chosen during scaffolding. This manifest is the single source of truth for ALL agents — Scaffolder creates it, every other agent reads it before working. Triggers on any scaffolding, implementation, review, or documentation task. Ensures cross-agent consistency for template patterns across the entire SDLC lifecycle.
Comprehensive product QA checklist for enterprise applications and AI agents, covering UX/accessibility, core functionality, LLM behavior, data handling, error resilience, security, performance, deployment hygiene, and observability. Use when performing full product QA, pre-release validation, or generating manual QA checklists. Triggers on product QA, project QA, release readiness, or comprehensive review requests. Complements the sdlc-qa-bug-checklist skill (which focuses on known bug patterns from ADO data).
Scaffold new application project structures from templates with CI/CD pipelines, Dockerfiles, and devcontainers. Use when creating a new service, restructuring a project, or setting up CI/CD. Triggers on scaffold, new project, new service, project structure, CI/CD pipeline, or Dockerfile requests.
Bug-driven QA checklist distilled from 338 real bugs across 9 CSACTOSOL ADO projects with detailed error patterns, Azure error codes, and verification commands. Use when reviewing code, PRs, deployments, or running QA checks. Triggers on any QA review, pre-deployment validation, or bug-prevention audit.
Manage the living reference catalog — research methodology, population rules, consumption rules, append-only enforcement, and review checkpoint behavior. Activated by the Analyst (population), Scaffolder/Implementer/Deployer/Documenter (consumption), and Harness (review checkpoint).
Guides the Analyst agent through collaborative requirements discovery with the user. Covers elicitation patterns, question strategies, restatement techniques, and the explicit approval checkpoint. Use when starting a new feature, clarifying requirements, or when Harness delegates Phase 1A discovery to the Analyst.
Structured YAML output format for SDLC QA reviewer agents. Ensures consistent, parseable review output across all 9 reviewer domains. Use in conjunction with each reviewer's existing Markdown report format.
Perform SDLC-aligned security review combining OWASP Top 10 with project-specific Azure patterns. Use when reviewing code for security vulnerabilities, secrets exposure, auth issues, injection risks, or OWASP compliance. Triggers on any code review, PR review, or security audit request. Loads fresh OWASP checklist from awesome-copilot MCP on every review.
Initialize a new repository with SDLC workspace files — MCP config, copilot-instructions.md, quality instructions, and prompt files. Use when setting up a new project, bootstrapping SDLC, onboarding a repo, or when Harness detects missing workspace files.