{"id":47594868,"url":"https://github.com/mcleo-d/claude-agents","last_synced_at":"2026-04-01T17:59:13.511Z","repository":{"id":344049318,"uuid":"1180241358","full_name":"mcleo-d/claude-agents","owner":"mcleo-d","description":"18 Claude Code subagent definitions forming a complete virtual engineering team","archived":false,"fork":false,"pushed_at":"2026-03-22T18:32:32.000Z","size":124,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-03-23T09:52:06.789Z","etag":null,"topics":["ai-agents","anthropic","claude","claude-code","subagents"],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/mcleo-d.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":".github/CODEOWNERS","security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-03-12T21:04:22.000Z","updated_at":"2026-03-22T18:32:35.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/mcleo-d/claude-agents","commit_stats":null,"previous_names":["mcleo-d/claude-agents"],"tags_count":2,"template":false,"template_full_name":null,"purl":"pkg:github/mcleo-d/claude-agents","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mcleo-d%2Fclaude-agents","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mcleo-d%2Fclaude-agents/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mcleo-d%2Fclaude-agents/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mcleo-d%2Fclaude-agents/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mcleo-d","download_url":"https://codeload.github.com/mcleo-d/claude-agents/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mcleo-d%2Fclaude-agents/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31290737,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-01T13:12:26.723Z","status":"ssl_error","status_checked_at":"2026-04-01T13:12:25.102Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-agents","anthropic","claude","claude-code","subagents"],"created_at":"2026-04-01T17:59:12.723Z","updated_at":"2026-04-01T17:59:13.495Z","avatar_url":"https://github.com/mcleo-d.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# claude-agents\n\n[![CI](https://github.com/mcleo-d/claude-agents/actions/workflows/ci.yml/badge.svg)](https://github.com/mcleo-d/claude-agents/actions/workflows/ci.yml)\n\nA collection of 18 Claude Code subagent definitions that form a complete virtual engineering team. Drop them into your `~/.claude/agents/` directory and Claude Code gains a bench of senior practitioners — each with deep domain expertise, clear accountability to the others, and a shared set of engineering principles.\n\n## What this is\n\nClaude Code supports [subagents](https://docs.anthropic.com/en/docs/claude-code/sub-agents): specialised AI personas defined in Markdown files that Claude can invoke automatically or on request. Each agent in this collection represents a senior role in a software delivery team — from business analyst through to SRE — and is designed to work alongside the others, not in isolation.\n\nThe collection covers the full delivery lifecycle:\n\n| Agent | Role | Invocation trigger |\n|---|---|---|\n| `business-analyst` | Requirements, user stories, BDD Gherkin, journey mapping | Capturing or refining requirements; writing acceptance criteria |\n| `systems-architect` | ADRs, C4 diagrams, technology selection, cross-cutting concerns | Significant design decisions; new services; API contracts |\n| `backend-developer` | Node.js/TypeScript, Go, CouchDB, REST and GraphQL APIs | Server-side services, background workers, data models |\n| `frontend-developer` | React/Next.js, TypeScript, accessibility, Core Web Vitals | UI components, pages, client-side state, browser TypeScript |\n| `fullstack-developer` | Vertical-slice feature delivery across database, API, and UI | Self-contained features spanning all layers |\n| `python-developer` | Python 3.9+, stdlib services, systemd logging, TDD | Python code — services, CLI tools, HTTP servers, automation |\n| `ai-ml-engineer` | Local/edge LLM deployment, Ollama, model selection and benchmarking | Selecting, tuning, or benchmarking LLMs on constrained hardware |\n| `ui-designer` | Design tokens, Tailwind config, component specs, accessibility standards | Design system, colour palette, typography, component specifications |\n| `qa-engineer` | Test pyramid, Playwright E2E, k6 load tests, Pact contracts | Test strategy, test suites, coverage standards |\n| `security-engineer` | Threat modelling, OWASP, STRIDE, IAM, DevSecOps pipeline | Security reviews, secrets management, IAM policies |\n| `devops-engineer` | GitHub Actions, AWS ECS/ECR, OpenTofu, Docker, ArgoCD | CI/CD pipelines, infrastructure, deployment strategies |\n| `platform-engineer` | Backstage, golden path templates, ArgoCD, Kong, Linkerd | Internal developer platform, self-service workflows |\n| `sre-engineer` | SLOs, error budgets, Prometheus, Grafana, on-call runbooks | Reliability targets, alerting, post-incident reviews |\n| `linux-systems-engineer` | Hardened bare-metal and edge Linux configuration | systemd, SSH hardening, UFW, fail2ban, ARM64 deployments |\n| `code-reviewer` | Multi-language PR review with structured severity framework | Reviewing any pull request or code change |\n| `systematic-debugger` | Hypothesis-driven fault diagnosis across the full stack | Bugs, production incidents, test failures |\n| `deploy-checklist` | Pre-deployment go/no-go validation for cloud and edge targets | Immediately before promoting a build to staging or production |\n| `scrum-master` | Sprint ceremonies, Definition of Done, retrospectives, DORA metrics | Sprint facilitation, process health, team impediments |\n\n## Installation\n\n```bash\n# Clone the repository\ngit clone https://github.com/mcleo-d/claude-agents.git\n\n# Copy all agents to your Claude Code agents directory\ncp claude-agents/agents/*.md ~/.claude/agents/\n```\n\nAfter copying, Claude Code will make the agents available immediately — no restart required. You can verify by running `/agents` in a Claude Code session.\n\n### Install a subset\n\nIf you only want specific agents, copy individual files:\n\n```bash\ncp claude-agents/agents/code-reviewer.md ~/.claude/agents/\ncp claude-agents/agents/security-engineer.md ~/.claude/agents/\n```\n\n### Keep up to date\n\n```bash\ncd claude-agents \u0026\u0026 git pull \u0026\u0026 cp agents/*.md ~/.claude/agents/\n```\n\n## Design philosophy\n\n### T-shaped agents\n\nEach agent is designed with a T-shape: deep expertise in its domain (the vertical bar) and genuine awareness of adjacent concerns (the horizontal bar). A backend developer who knows nothing about accessibility or the security engineer's authority model is not a senior engineer.\n\n### Collaborative by design\n\nAgents are not standalone — they coordinate. Every agent declares an explicit `## Interaction model` that names which other agents it hands work to and receives work from. The relationships are bidirectional: if `qa-engineer` references `backend-developer`, `backend-developer` references `qa-engineer` back.\n\n### Security authority model\n\nThe `security-engineer` is the explicit authority on security design decisions across the entire team. Every agent that touches security concerns declares deference: it may identify and escalate, but it does not design security controls unilaterally. This prevents security fragmentation without creating a bottleneck.\n\n### Five shared principles\n\nAll agents share five core principles, each adapted to the specific concerns of their domain:\n\n1. **Customer orientation** — every decision traces back to user impact\n2. **Accessibility first** — accessibility is a first-class quality gate, not an afterthought\n3. **Ethical engineering and diversity of thought** — challenge monoculture assumptions; surface bias\n4. **Environmental and cost responsibility** — right-size compute, data, and infrastructure choices\n5. **Test-driven development** — evidence and verification over assumption\n\n### Model tiers\n\nAgents that make high-stakes, nuanced judgements run on `claude-opus-4-6`:\n\n- `code-reviewer` — final quality gate before merge\n- `security-engineer` — security design authority\n- `systems-architect` — architectural decision authority\n- `systematic-debugger` — complex, multi-hypothesis fault diagnosis\n\nAll other agents run on `claude-sonnet-4-6`.\n\n## Customising for your stack\n\nThe agents are intentionally stack-agnostic within their domain. You will likely want to adapt:\n\n- **Technology choices** — swap CouchDB for PostgreSQL in `backend-developer`, or replace AWS with GCP in `devops-engineer`\n- **Tool names** — update specific tooling references (e.g., Playwright → Cypress in `qa-engineer`)\n- **Interaction model** — add or remove agent relationships to match your actual team structure\n- **Coverage gates** — adjust percentage thresholds in `qa-engineer` to match your project's standards\n\nWhen adapting, preserve the five core principles, the interaction model section, and the security deference model. These are the structural elements that make the agents work as a team.\n\n## Contributing\n\nContributions are welcome. See [CONTRIBUTING.md](CONTRIBUTING.md) for the agent structure requirements, quality bar, and PR checklist.\n\nUse the issue templates to propose a new agent or suggest improvements to an existing one.\n\n## Licence\n\nCopyright 2026 James McLeod\n\nLicensed under the Apache License, Version 2.0. See [LICENSE](LICENSE) for the full text.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmcleo-d%2Fclaude-agents","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmcleo-d%2Fclaude-agents","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmcleo-d%2Fclaude-agents/lists"}