SWE Marketplace

A curated Claude Code plugin marketplace for practical, everyday usage in software engineering — 12 plugins, 53 specialist agents, 10 skills, 3 commands. A few opinionated choices that set it apart from larger awesome-style lists:

• Curated, not exhaustive. Every plugin and agent is one I actually use. No vendored grab-bags, no auto-generated stubs of skills the model already has.
• SDK-loadable. Every plugin ships both the catalog entry and a per-plugin plugin.json, so consumers using the Claude Agent SDK (ClaudeAgentOptions.plugins=[{"type":"local","path":...}] or --plugin-dir) can load them directly. Many marketplaces only ship the catalog and silently fail to register skills/commands/hooks when loaded that way.
• CI-guarded drift. A GitHub Action validates manifest sync on every PR.
• Zero-warning validation. Every plugin passes claude plugin validate with no errors or warnings. = Most skills & plugins have been run through exhaustive evals. I am an avid fan of skill-creator and have a custom-build eval harness for building and evaluating plugins. Many of the resources here are fruit borne from that work.

Quick start

# In Claude Code
/plugin marketplace add https://github.com/andisab/swe-marketplace.git
/plugin install context-engineering@swe-marketplace
# Restart Claude Code, then verify with /help or /agents

Browse the full catalog with /plugin marketplace list. Every plugin is claude plugin validate-clean and version-pinned in marketplace.json.

What's inside

adv — Adversarial multi-model code review (v0.2.0)

Cross-examines code changes using Claude, Codex CLI, and Gemini CLI together. One agent orchestrates 5 specialized reviewers; three commands let you talk to each model directly.

• Agent: adv-review
• Skill: dispatch (shared multi-model CLI infrastructure)
• Commands: /adv-codex, /adv-gemini, /adv-gemini-research

context-engineering — The plugin you use to build other plugins (v1.1.0)

End-to-end toolkit for authoring Claude Code resources following Anthropic specs. Templates, examples, and discovery-optimized patterns for every component type.

• Agent: context-engineer
• Skills: agent-dev, skill-dev, command-dev, hook-dev, plugin-dev, mcp-tool-dev, mcp-server-dev

research-team — Multi-agent research orchestration (v1.2.3)

A coordinator skill decomposes a topic into subtopics, spawns parallel research-specialist subagents, then hands off to a research-report-writer for synthesis. Joplin output supported.

• Agents: research-specialist, research-report-writer
• Skills: coordinator, joplin-research

dev — 13 language experts (v1.0.0)

frontend — UI and design specialists (v1.0.0)

• fe-react-expert, fe-vue-expert, fe-nextjs-expert
• fe-html-expert, fe-css-expert
• fe-frontend-designer (design systems, UX)

infrastructure — DevOps, cloud, and platform (v1.0.0)

• Cloud: infra-aws-architect, infra-gcp-architect
• Containers: infra-docker-engineer, infra-k8s-engineer
• IaC: infra-terraform-engineer
• CI/CD: infra-github-actions-expert, infra-gitlab-ci-expert
• Security: infra-security-auditor

databases — SQL, NoSQL, graph, and vector (v1.0.0)

db-postgres-expert, db-mongodb-expert, db-neo4j-expert, db-cassandra-expert, db-mariadb-expert, db-sql-expert, db-vector-expert

data — Data science and data engineering (v1.0.0)

• Visualization: data-d3-expert, data-highcharts-expert
• Notebooks: data-jupyter-expert, data-google-colab-expert
• Languages: data-r-expert, data-python-data-engineer

machine-learning — ML framework specialists (v1.0.0)

ml-pytorch-expert, ml-tensorflow-expert, ml-scikit-learn-expert

genai — Generative AI development (v1.0.0)

genai-langchain-expert — chains, agents, retrieval workflows. More frameworks on the way.

product-management — PRDs, content, planning (v1.0.0)

pm-prd-writer, pm-content-writer, pm-project-planner

arch — System and build orchestration (v1.0.0)

arch-context-agent (context management for large codebases), build-orchestrator (multi-agent build coordination)

Plugin manifests

Two manifests ship together:

• .claude-plugin/marketplace.json (repo root) — catalog used by /plugin marketplace add and /plugin install. Source of truth.
• plugins/<name>/.claude-plugin/plugin.json — per-plugin manifest used by SDK consumers loading plugins directly. Generated from the matching marketplace.json entry.

python scripts/synthesize-plugin-manifests.py          # regenerate
python scripts/synthesize-plugin-manifests.py --check  # drift check (CI runs this)
./scripts/validate-all.sh                              # full validation

See CONTRIBUTING.md for the full contribution workflow.

Other marketplaces worth checking out

Safety note: verify the source of any marketplace before adding it. Plugins can register hooks and MCP servers that run on your machine — read what you're running and review permissions carefully.

Documentation

Plugins guide · Plugin reference · Marketplaces · Sub-agents · Skills · Slash commands · Hooks

Contributing

Contributions welcome — agent tweaks, new plugins, doc fixes, bug reports. See CONTRIBUTING.md.

License

MIT — © Andis A. Blukis.

Contact

• GitHub: @andisab
• LinkedIn: andisab
• Email: [email protected]