Agent Brain Plugin Plan

Implementation Plan: Agent Brain Plugin Conversion

Branch: 114-agent-brain-plugin | Date: 2025-01-31 | Spec: spec.md Input: Feature specification from .speckit/features/114-agent-brain-plugin/spec.md

Summary

Convert the existing Agent Brain skill into a full Claude Code plugin with marketplace.json format. The plugin provides 15 slash commands for search, setup, server management, and indexing operations. It includes 2 agents for proactive assistance and 2 skills with reference documentation. The implementation uses the creating-plugin-from-skill workflow and achieves 95% quality score through the improving-skills process.

Technical Context

Language/Version: Markdown (Claude Code plugin format) Primary Dependencies: Claude Code plugin system, agent-brain-cli v1.2.0+, agent-brain-rag v1.2.0+ Storage: N/A (plugin is markdown files only) Testing: Manual plugin installation and command execution Target Platform: Claude Code (any OS with Claude Code installed) Project Type: Claude Code plugin (markdown-based) Performance Goals: Commands execute within 3 seconds for typical queries Constraints: Must follow marketplace.json schema, skill grading must reach 95% Scale/Scope: 15 commands, 2 agents, 2 skills, ~25 markdown files total

Constitution Check

GATE: Must pass before Phase 0 research. Re-check after Phase 1 design.

Principle	Status	Notes
I. Monorepo Modularity	✅ PASS	Plugin is a new independent package (agent-brain-plugin) that consumes server/CLI APIs
II. OpenAPI-First	✅ N/A	Plugin consumes existing OpenAPI endpoints, does not define new ones
III. Test-Alongside	✅ PASS	Plugin quality validated through skill grading (95% target)
IV. Observability	✅ PASS	Plugin commands use existing /health and /status endpoints
V. Simplicity	✅ PASS	Uses standard plugin format, no custom abstractions

Gate Status: PASSED - All applicable principles satisfied.

Project Structure

Documentation (this feature)

.speckit/features/114-agent-brain-plugin/
├── plan.md              # This file
├── spec.md              # Feature specification
├── research.md          # Phase 0 output
├── data-model.md        # Phase 1 output (plugin structure model)
├── quickstart.md        # Phase 1 output
├── contracts/           # Phase 1 output (command/agent schemas)
└── tasks.md             # Phase 2 output (/speckit.tasks command)

Source Code (repository root)

agent-brain-plugin/
├── .claude-plugin/
│   └── marketplace.json       # Plugin manifest
├── commands/                   # 15 slash command files
│   ├── agent-brain-search.md
│   ├── agent-brain-semantic.md
│   ├── agent-brain-keyword.md
│   ├── agent-brain-install.md
│   ├── agent-brain-setup.md
│   ├── agent-brain-config.md
│   ├── agent-brain-init.md
│   ├── agent-brain-verify.md
│   ├── agent-brain-start.md
│   ├── agent-brain-stop.md
│   ├── agent-brain-status.md
│   ├── agent-brain-list.md
│   ├── agent-brain-index.md
│   ├── agent-brain-reset.md
│   └── agent-brain-help.md
├── agents/                     # 2 agent files
│   ├── search-assistant.md
│   └── setup-assistant.md
├── skills/                     # 2 skills with references
│   ├── using-agent-brain/
│   │   ├── SKILL.md
│   │   └── references/
│   │       ├── hybrid-search-guide.md
│   │       ├── bm25-search-guide.md
│   │       ├── vector-search-guide.md
│   │       └── api_reference.md
│   └── agent-brain-setup/
│       ├── SKILL.md
│       └── references/
│           ├── installation-guide.md
│           ├── configuration-guide.md
│           └── troubleshooting-guide.md
├── README.md
└── .gitignore

Structure Decision: Claude Code Plugin structure following marketplace.json format. All files are markdown (commands, agents, skills) with YAML frontmatter for metadata.

Complexity Tracking

No violations - plugin uses standard Claude Code plugin format with no custom abstractions.