Agent Brain Plugin Tasks

Tasks: Agent Brain Plugin Conversion

Input: Design documents from .speckit/features/114-agent-brain-plugin/ Prerequisites: plan.md, spec.md, research.md, data-model.md, contracts/

Tests: Not explicitly requested - quality validation via skill grading workflow (95% target)

Organization: Tasks grouped by user story to enable independent implementation and testing

Format: [ID] [P?] [Story] Description

• [P]: Can run in parallel (different files, no dependencies)
• [Story]: Which user story this task belongs to (e.g., US1, US2, US3)
• Include exact file paths in descriptions

Path Conventions

• Plugin root: agent-brain-plugin/ at repository root
• Commands: agent-brain-plugin/commands/
• Agents: agent-brain-plugin/agents/
• Skills: agent-brain-plugin/skills/

---

Phase 1: Setup (Plugin Structure)

Purpose: Create plugin directory structure and manifest

• T001 Create plugin directory structure: agent-brain-plugin/ with .claude-plugin/, commands/, agents/, skills/ subdirectories (IN PROGRESS - Claude Code Task #5)
• T002 Create marketplace.json manifest in agent-brain-plugin/.claude-plugin/marketplace.json per data-model.md schema (IN PROGRESS - Claude Code Task #5)
• T003 [P] Create README.md in agent-brain-plugin/README.md with installation and usage instructions (IN PROGRESS - Claude Code Task #5)
• T004 [P] Create .gitignore in agent-brain-plugin/.gitignore for common excludes (IN PROGRESS - Claude Code Task #5)

Checkpoint: Plugin structure ready for content

---

Phase 2: Foundational (Skills Infrastructure)

Purpose: Create skill foundations that commands and agents depend on

⚠️ CRITICAL: Commands and agents reference skills - skills must exist first

using-agent-brain Skill

• T005 Create skill directory structure: agent-brain-plugin/skills/using-agent-brain/ with references/ subdirectory (IN PROGRESS - Claude Code Task #5)
• T006 Create SKILL.md in agent-brain-plugin/skills/using-agent-brain/SKILL.md per skill-schema.md (IN PROGRESS - Claude Code Task #5)
• T007 [P] Create hybrid-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/hybrid-search-guide.md (IN PROGRESS - Claude Code Task #5)
• T008 [P] Create bm25-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/bm25-search-guide.md (IN PROGRESS - Claude Code Task #5)
• T009 [P] Create vector-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/vector-search-guide.md (IN PROGRESS - Claude Code Task #5)
• T010 [P] Create api_reference.md in agent-brain-plugin/skills/using-agent-brain/references/api_reference.md (IN PROGRESS - Claude Code Task #5)

agent-brain-setup Skill

• T011 Create skill directory structure: agent-brain-plugin/skills/agent-brain-setup/ with references/ subdirectory (IN PROGRESS - Claude Code Task #5)
• T012 Create SKILL.md in agent-brain-plugin/skills/agent-brain-setup/SKILL.md per skill-schema.md (IN PROGRESS - Claude Code Task #5)
• T013 [P] Create installation-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/installation-guide.md (IN PROGRESS - Claude Code Task #5)
• T014 [P] Create configuration-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/configuration-guide.md (IN PROGRESS - Claude Code Task #5)
• T015 [P] Create troubleshooting-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/troubleshooting-guide.md (IN PROGRESS - Claude Code Task #5)

Checkpoint: Skills ready - commands and agents can now reference them

---

Phase 3: User Story 1 - Quick Search with Plugin Commands (Priority: P1) 🎯 MVP

Goal: Users can search project documentation using natural commands

Independent Test: Install plugin, start server, index docs, run /agent-brain-search "query" and verify results with source citations

Search Commands

• T016 [P] [US1] Create agent-brain-search.md command in agent-brain-plugin/commands/agent-brain-search.md per command-schema.md (hybrid mode)
• T017 [P] [US1] Create agent-brain-semantic.md command in agent-brain-plugin/commands/agent-brain-semantic.md per command-schema.md (vector mode)
• T018 [P] [US1] Create agent-brain-keyword.md command in agent-brain-plugin/commands/agent-brain-keyword.md per command-schema.md (BM25 mode)

Supporting Commands for Search

• T019 [P] [US1] Create agent-brain-status.md command in agent-brain-plugin/commands/agent-brain-status.md (needed to check server before search)
• T020 [P] [US1] Create agent-brain-start.md command in agent-brain-plugin/commands/agent-brain-start.md (needed if server not running)

Search Agent

• T021 [US1] Create search-assistant.md agent in agent-brain-plugin/agents/search-assistant.md per agent-schema.md

Checkpoint: User Story 1 complete - users can search documentation via commands

---

Phase 4: User Story 2 - Guided Installation and Setup (Priority: P2)

Goal: New users can install and configure Agent Brain from scratch

Independent Test: Fresh environment, run /agent-brain-setup, follow guided flow to completion, verify with /agent-brain-verify

Setup Commands

• T022 [P] [US2] Create agent-brain-install.md command in agent-brain-plugin/commands/agent-brain-install.md per command-schema.md
• T023 [P] [US2] Create agent-brain-setup.md command in agent-brain-plugin/commands/agent-brain-setup.md (guided flow)
• T024 [P] [US2] Create agent-brain-config.md command in agent-brain-plugin/commands/agent-brain-config.md (API key configuration)
• T025 [P] [US2] Create agent-brain-init.md command in agent-brain-plugin/commands/agent-brain-init.md (project initialization)
• T026 [P] [US2] Create agent-brain-verify.md command in agent-brain-plugin/commands/agent-brain-verify.md (installation verification)

Setup Agent

• T027 [US2] Create setup-assistant.md agent in agent-brain-plugin/agents/setup-assistant.md per agent-schema.md

Checkpoint: User Story 2 complete - new users can set up Agent Brain via guided commands

---

Phase 5: User Story 3 - Server Lifecycle Management (Priority: P2)

Goal: Users can manage the Agent Brain server for their project

Independent Test: Run /agent-brain-start, verify with /agent-brain-status, run /agent-brain-list to see instance, run /agent-brain-stop

Server Commands

• T028 [P] [US3] Create agent-brain-stop.md command in agent-brain-plugin/commands/agent-brain-stop.md per command-schema.md
• T029 [P] [US3] Create agent-brain-list.md command in agent-brain-plugin/commands/agent-brain-list.md (list all instances)

Note: agent-brain-start.md and agent-brain-status.md already created in US1 (T019, T020)

Checkpoint: User Story 3 complete - users can manage server lifecycle

---

Phase 6: User Story 4 - Document Indexing via Commands (Priority: P2)

Goal: Users can index their project documentation for search

Independent Test: Run /agent-brain-index /docs, check status shows document count, run /agent-brain-reset and confirm index cleared

Indexing Commands

• T030 [P] [US4] Create agent-brain-index.md command in agent-brain-plugin/commands/agent-brain-index.md per command-schema.md
• T031 [P] [US4] Create agent-brain-reset.md command in agent-brain-plugin/commands/agent-brain-reset.md (clear index with confirmation)

Checkpoint: User Story 4 complete - users can index and reset documentation

---

Phase 7: User Story 5 - Proactive Agent Assistance (Priority: P2)

Goal: Agents proactively offer help when user mentions search or setup topics

Independent Test: Say "search docs" and verify agent offers to help; say "how do I search" and verify agent explains Agent Brain

Note: Both agents already created in US1 (T021) and US2 (T027)

• T032 [US5] Review and enhance search-assistant.md trigger patterns for comprehensive coverage in agent-brain-plugin/agents/search-assistant.md
• T033 [US5] Review and enhance setup-assistant.md trigger patterns for comprehensive coverage in agent-brain-plugin/agents/setup-assistant.md

Checkpoint: User Story 5 complete - agents trigger appropriately on patterns

---

Phase 8: User Story 6 - Interactive Help and Guidance (Priority: P3)

Goal: Users can get help with commands and search mode selection

Independent Test: Run /agent-brain-help and verify all 15 commands listed; run /agent-brain-help --command search for detailed help

Help Command

• T034 [US6] Create agent-brain-help.md command in agent-brain-plugin/commands/agent-brain-help.md per command-schema.md

Checkpoint: User Story 6 complete - users have comprehensive help available

---

Phase 9: Quality Assurance (Required - 95% Target)

Purpose: Achieve 95% skill grading score for both skills

Skill Grading

• T035 Grade using-agent-brain skill with improving-skills workflow
• T036 Identify and fix issues in agent-brain-plugin/skills/using-agent-brain/SKILL.md to improve score
• T037 Re-grade using-agent-brain skill until 95% achieved
• T038 Grade agent-brain-setup skill with improving-skills workflow
• T039 Identify and fix issues in agent-brain-plugin/skills/agent-brain-setup/SKILL.md to improve score
• T040 Re-grade agent-brain-setup skill until 95% achieved

Checkpoint: Both skills achieve 95% quality score

---

Phase 10: Polish & Cross-Cutting Concerns

Purpose: Final validation and publication preparation

• T041 Update marketplace.json to ensure all commands, agents, skills are correctly referenced in agent-brain-plugin/.claude-plugin/marketplace.json
• T042 [P] Validate all command YAML frontmatter is correctly formatted
• T043 [P] Validate all agent trigger patterns work as expected
• T044 [P] Validate all skill references load correctly
• T045 Plugin validated as part of monorepo (no separate installation needed)
• T046 N/A - Plugin is part of existing agent-brain monorepo
• T047 N/A - Will be committed with main repository
• T048 N/A - Plugin path documented in monorepo README

---

Dependencies & Execution Order

Phase Dependencies

• Setup (Phase 1): No dependencies - can start immediately
• Foundational (Phase 2): Depends on Setup - BLOCKS all user story commands
• User Stories (Phase 3-8): All depend on Foundational phase completion
• Quality Assurance (Phase 9): Depends on all skills being created (Phase 2)
• Polish (Phase 10): Depends on all content being complete

User Story Dependencies

Story	Priority	Dependencies	Can Parallel With
US1 - Quick Search	P1	Foundational	None (MVP)
US2 - Guided Setup	P2	Foundational	US1, US3, US4, US6
US3 - Server Lifecycle	P2	Foundational, US1 (T019, T020)	US2, US4, US6
US4 - Document Indexing	P2	Foundational	US1, US2, US3, US6
US5 - Agent Assistance	P2	US1 (T021), US2 (T027)	None
US6 - Help & Guidance	P3	Foundational	US1, US2, US3, US4

Within Each User Story

• Commands marked [P] can run in parallel (different files)
• Agent files depend on having skill references available
• Each story should be testable after completion

Parallel Opportunities

Phase 1 (Setup):

• T003, T004 can run in parallel

Phase 2 (Foundational):

• T007, T008, T009, T010 can run in parallel (using-agent-brain references)
• T013, T014, T015 can run in parallel (agent-brain-setup references)

User Stories:

• All commands within each story marked [P] can run in parallel
• US1, US2, US3, US4, US6 can run in parallel after Foundational
• US5 must wait for agents from US1/US2

---

Parallel Example: Phase 2 (Foundational)

# Launch all using-agent-brain reference files together:
Task: "Create hybrid-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/hybrid-search-guide.md"
Task: "Create bm25-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/bm25-search-guide.md"
Task: "Create vector-search-guide.md in agent-brain-plugin/skills/using-agent-brain/references/vector-search-guide.md"
Task: "Create api_reference.md in agent-brain-plugin/skills/using-agent-brain/references/api_reference.md"

# Launch all agent-brain-setup reference files together:
Task: "Create installation-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/installation-guide.md"
Task: "Create configuration-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/configuration-guide.md"
Task: "Create troubleshooting-guide.md in agent-brain-plugin/skills/agent-brain-setup/references/troubleshooting-guide.md"

Parallel Example: User Story 1 (Search Commands)

# Launch all search commands together:
Task: "Create agent-brain-search.md command in agent-brain-plugin/commands/agent-brain-search.md"
Task: "Create agent-brain-semantic.md command in agent-brain-plugin/commands/agent-brain-semantic.md"
Task: "Create agent-brain-keyword.md command in agent-brain-plugin/commands/agent-brain-keyword.md"
Task: "Create agent-brain-status.md command in agent-brain-plugin/commands/agent-brain-status.md"
Task: "Create agent-brain-start.md command in agent-brain-plugin/commands/agent-brain-start.md"

---

Implementation Strategy

MVP First (User Story 1 Only)

1. Complete Phase 1: Setup (T001-T004)
2. Complete Phase 2: Foundational - skills (T005-T015)
3. Complete Phase 3: User Story 1 - search commands (T016-T021)
4. STOP and VALIDATE: Install plugin, test search commands work
5. Deploy/demo if ready

Incremental Delivery

1. Setup + Foundational → Plugin structure ready
2. Add User Story 1 (Search) → Test independently → MVP Ready!
3. Add User Story 2 (Setup) → Test independently → First-time user support
4. Add User Story 3 (Server) → Test independently → Full server lifecycle
5. Add User Story 4 (Indexing) → Test independently → Complete indexing
6. Add User Story 5 (Agents) → Test independently → Proactive assistance
7. Add User Story 6 (Help) → Test independently → Complete help system
8. Quality Assurance → 95% skill scores achieved
9. Polish → GitHub publication

Parallel Team Strategy

With multiple contributors:

1. Everyone completes Setup + Foundational together
2. Once Foundational is done:
   • Contributor A: US1 (Search) + US5 (Agents)
   • Contributor B: US2 (Setup) + US6 (Help)
   • Contributor C: US3 (Server) + US4 (Indexing)
3. All run Quality Assurance and Polish together

---

Task Summary

Phase	Task Count	Parallel Tasks	Description
1: Setup	4	2	Plugin structure
2: Foundational	11	7	Skills infrastructure
3: US1 Search	6	5	Search commands + agent
4: US2 Setup	6	5	Setup commands + agent
5: US3 Server	2	2	Server lifecycle commands
6: US4 Indexing	2	2	Index commands
7: US5 Agents	2	0	Agent pattern review
8: US6 Help	1	0	Help command
9: Quality	6	0	95% skill grading
10: Polish	8	3	Validation & publication
Total	48	26

---

Notes

• [P] tasks = different files, no dependencies
• [Story] label maps task to specific user story for traceability
• Each user story should be independently completable and testable
• Quality assurance (95% grading) is mandatory per spec
• Commit after each task or logical group
• Stop at any checkpoint to validate story independently
• Existing skill content from agent-brain-skill/using-agent-brain/ can be migrated