{"id":"5ce8400b-10df-45b2-a376-8c1cf131c5be","slug":"claude-plugin-software-engineering-team","kind":"skill","name":"claude-plugin-software-engineering-team","summary":"Opinionated specialist agents and team workflows for Claude Code. Review, investigate, and spike with parallel multi-agent teams.","owner":"rayh","repoUrl":"https://github.com/rayh/claude-plugin-software-engineering-team","license":"MIT","licenseTier":"permissive","trust":"community","stars":0,"match":null,"fewStars":true,"format":"skill","createdAt":"2026-02-18T07:00:05.000Z","addedAt":"2026-10-22T02:22:58.148Z","logoUrl":"https://avatars.githubusercontent.com/u/57468?u=123a1cc66e1972e0baf95f0ed3f712af0de3745e&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["developer-tools","communication"],"tags":[],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-23T22:35:04.969Z","sample":false,"url":"https://conduits.dev/tools/claude-plugin-software-engineering-team","language":"","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# claude-plugin-software-engineering-team\n\nOpinionated specialist agents and team workflows for Claude Code. Focused on doing fewer things well rather than covering everything.\n\n## Philosophy\n\nInspired by [Trail of Bits' approach](https://github.com/trailofbits/claude-code-config): agents encode **expertise, not procedures**. Each agent knows *how to think* about its domain, not just what steps to follow. They include reference material (anti-pattern catalogues, checklists, smell guides) that make them expert-level rather than superficial.\n\nKey principles:\n- **Agents don't overlap** — each has a clear domain boundary and explicitly states what it doesn't do\n- **Severity is standardised** — Critical/Warning/Suggestion across all agents, with consistent meaning\n- **Memory is enabled** — agents learn your codebase patterns across sessions via `memory: project`\n- **Cost-conscious** — review agents use Sonnet (fast, cheap for read-only analysis); only the orchestrator inherits your session model\n\n## What's Included\n\n### Review Agents\n\n| Agent | Domain | Key Focus |\n|-------|--------|-----------|\n| `software-architect` | Code quality | Design patterns, SOLID, coupling, cohesion, naming, dead code |\n| `sre-architect` | Operations | Cost, performance, reliability, observability, failure modes |\n| `quality-engineer` | Testing | Behavioural coverage, test quality, strategy, testability, risk |\n| `security-auditor` | App security | OWASP top 10, injection, auth, data exposure, dependencies |\n\n### Orchestration\n\n| Agent | Role |\n|-------|------|\n| `tech-lead` | Reads your changes, decides which specialists to involve, synthesises findings |\n\n### Investigation\n\n| Agent | Role |\n|-------|------|\n| `investigator` | Pursues a single hypothesis. Spawned in teams for competing-hypotheses debugging |\n\n### Skills (Slash Commands)\n\n| Command | What It Does |\n|---------|-------------|\n| `/review [scope]` | Spawns architect + SRE + quality in parallel. Scope can be a path, branch, or \"staged\" |\n| `/investigate [bug description]` | Spawns 3-5 investigators each pursuing different hypotheses, then debate |\n| `/spike [question]` | Spawns one agent per competing approach to evaluate trade-offs side by side |\n\n## Installation\n\n### Option 1: Claude Code Plugin (recommended)\n\n```bash\n# From any Claude Code session:\n/plugin install https://github.com/rayh/claude-plugin-software-engineering-team\n```\n\n### Option 2: Copy to Global Config\n\n```bash\n# Clone the repo\ngit clone https://github.com/rayh/claude-plugin-software-engineering-team.git\ncd claude-plugin-software-engineering-team\n\n# Copy agents and skills to your global Claude config\ncp -r agents/* ~/.claude/agents/\nmkdir -p ~/.claude/skills/review ~/.claude/skills/investigate ~/.claude/skills/spike\ncp skills/review/SKILL.md ~/.claude/skills/review/\ncp skills/investigate/SKILL.md ~/.claude/skills/investigate/\ncp skills/spike/SKILL.md ~/.claude/skills/spike/\n```\n\n### Option 3: Copy to a Specific Project\n\n```bash\n# From your project root\ncp -r agents/* .claude/agents/\ncp -r skills/* .claude/skills/\n```\n\n### Enable Agent Teams\n\nAgent teams are experimental. Add to your `~/.claude/settings.json`:\n\n```json\n{\n  \"env\": {\n    \"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS\": \"1\"\n  }\n}\n```\n\n## Usage\n\n### Quick Review (After Making Changes)\n\n```\n/review staged\n/review src/api/\n/review feature-branch\n```\n\n### Smart Review (Let Tech Lead Decide)\n\n```\nUse the tech-lead to review my recent changes\n```\n\nThe tech-lead reads the diff, consults its decision matrix, and only spawns relevant specialists. A one-line bug fix gets quality-engineer only. An API change gets all four.\n\n### Bug Investigation\n\n```\n/investigate Users report intermittent 500 errors on POST /api/sessions after deploying v2.3\n```\n\n### Technical Spike\n\n```\n/spike Should we use Redis, DynamoDB, or PostgreSQL for session storage?\n```\n\n### Single Specialist\n\n```\nUse the security-auditor to review the OAuth implementation in src/auth/\nUse the sre-architect to evaluate the docker-compose.yml\n```\n\n## When to Use What\n\n| Situation | Use |\n|-----------|-----|\n| Finished a feature, want review | `/review` |\n| Small fix, just want test review | `Use quality-engineer to review...` |\n| Infra change | `Use sre-architect to review...` |\n| Weird bug, unclear root cause | `/investigate` |\n| Choosing between technologies | `/spike` |\n| Complex change, need orchestration | `Use tech-lead to review...` |\n| Security-sensitive change | `Use security-auditor to audit...` |\n\n## Team Topology Ideas\n\nBeyond the included agents, these patterns work well with agent teams:\n\n### Migration/Refactor Team\n- **Scout**: maps all usages of the thing being changed\n- **Implementer(s)**: one per module to avoid file conflicts\n- **Verifier**: runs tests after each batch, reports regressions\n\n### Documentation Team\n- **Reader**: traces code flows, builds understanding\n- **Writer**: produces docs from reader's findings\n- **Reviewer**: checks docs against actual code\n\n### Incident Response Team\n- **Timeline builder**: reconstructs what happened from logs/metrics\n- **Root cause analyst**: investigates code for the underlying bug\n- **Remediation planner**: proposes fix + prevention measures\n\n## Customisation\n\nAll agents are markdown files with YAML frontmatter. Edit them to:\n\n- **Adjust severity calibration** — if an agent flags too many Warnings for things you consider fine\n- **Add project-specific context** — domain knowledge, tech stack details, team conventions\n- **Change models** — upgrade to `opus` for more thorough reviews, or `haiku` for quick checks\n- **Restrict tools** — remove `Bash` if you want purely read-only analysis\n\n## Prior Art & Inspiration\n\nThis project builds on ideas from:\n- [trailofbits/claude-code-config](https://github.com/trailofbits/claude-code-config) — \"encode expertise not procedures\"\n- [VoltAgent/awesome-claude-code-subagents](https://github.com/VoltAgent/awesome-claude-code-subagents) — 100+ agent definitions\n- [contains-studio/agents](https://github.com/contains-studio/agents) — departmental agent organisation\n- [affaan-m/everything-claude-code](https://github.com/affaan-m/everything-claude-code) — battle-tested config collection\n- [wshobson/agents](https://github.com/wshobson/agents) — plugin-based orchestration patterns\n- [Addy Osmani's agent teams writeup](https://addyosmani.com/blog/claude-code-agent-teams/) — narrow scope = better reasoning\n\n## License\n\nMIT\n"}