What is this?
SoloBoard is a Claude Code plugin that silently tracks your tasks as you work. No setup commands, no context switching, no overhead. Just code — the board manages itself.
you > fix the login bug on mobile
← task silently created in DOING
claude > Found the issue in auth.ts:47...
← files auto-tracked to task
you > commit this
← commit SHA auto-linked to task
you > /soloboard-board
TODO (1) DOING (1) DONE (3)
→ Fix login bug ✓ Setup auth
on mobile ✓ Add dark mode
abc1234 ✓ Refactor API
Install
Option A: npm global (recommended)
npm install -g soloboard
soloboard install # in your project directory
Option B: Clone & build
git clone https://github.com/egorfedorov/Soloboard.git
cd Soloboard && npm install && npm run build
bash install.sh /path/to/your/project
Then just cd your-project && claude — the board manages itself.
Getting Started
See the Getting Started Guide for a complete walkthrough — installation, first use, common workflows, and troubleshooting.
How it works
| You do | SoloBoard does (silently) |
|---|
"fix the login bug" | Creates task → DOING |
"how does auth work?" | Nothing — it's a question |
"add dark mode" | Moves previous task → TODO, creates new → DOING |
| Edit files | Auto-tracks changed files to active task |
git commit | Auto-links commit SHA to active task |
/soloboard-board | Shows the kanban board |
/soloboard-task done | Moves active task → DONE |
Commands
| Command | What it does |
|---|
/soloboard-board | View your kanban board |
/soloboard-task | Show active task / done / create <title> / delete <name> |
/soloboard-project | Show project / create <name> / list / switch <name> |
Features
- Zero friction — no setup commands, board auto-initializes on first prompt
- Silent tracking — tasks created from actionable prompts, questions ignored
- Smart task creation — analyzes project to find related files, auto-tags, and sets priority
- Task context — saves what you examined, decisions made, remaining work per task
- Task agents — auto-generates
.claude/agents/ files for complex tasks
- Auto-review — pre-close analysis checks for TODOs, tests, type errors before marking done
- Task splitting — break complex tasks into subtasks with progress tracking
- Dependencies — blocked-by relationships with circular dependency detection
- Critical path — find the bottleneck chain that determines project duration
- Sprints — time-boxed task grouping with burndown tracking
- Daily standup — automated standup summary: done, in-progress, blocked
- Pomodoro timer — focus sessions tied to tasks
- Auto-manager — project health score, stall detection, smart suggestions
- Gantt chart — text-based timeline view with dependency visualization
- Auto-reprioritize — smart priority adjustment based on blockers and progress
- Git integration — commits, branches, and PRs auto-linked to tasks
- Fuzzy search — say "move login bug to done" and it finds the right task
- File-per-task storage — each task is a JSON file in
.kanban/tasks/, git-friendly
- Safe — only touches
.kanban/ in your project, no network, no dangerous ops
- 3 statuses — TODO, DOING, DONE. That's it.
- Time tracking — automatic: tracks time in DOING, shows per-task and total
- Priority sorting — high tasks float to top of each column automatically
- Markdown export — export the board as markdown for reports or sharing
- Multi-project dashboard — see all projects at a glance with
dashboard
- Global CLI —
npm install -g soloboard && soloboard install
- Multi-agent orchestration — register agents, claim tasks, lock files, handoff context
Architecture
soloboard/
├── src/mcp-server/
│ ├── index.ts # Entry point (stdio transport)
│ ├── server.ts # MCP server + 94 tools
│ ├── tools/
│ │ ├── task-tools.ts # create/update/get/list/move/delete
│ │ ├── board-tools.ts # view/project-create/list/switch
│ │ ├── session-tools.ts # log/summary
│ │ ├── git-tools.ts # link/status
│ │ ├── init-tools.ts # auto_init/board_summary
│ │ ├── export-tools.ts # export/dashboard/prioritize/time
│ │ ├── smart-tools.ts # smart_create/analyze (project analysis)
│ │ ├── context-tools.ts # context_save/load (task continuity)
│ │ ├── agent-tools.ts # agent_create/delete (.claude/agents/)
│ │ ├── review-tools.ts # pre-close review & checklist
│ │ ├── dependency-tools.ts # depend/blockers/critical_path
│ │ ├── subtask-tools.ts # split/subtasks
│ │ ├── sprint-tools.ts # create/add/close/view
│ │ ├── standup-tools.ts # standup/pomodoro
│ │ ├── manager-tools.ts # report/stall/suggest/reprioritize/gantt
│ │ ├── orchestration-tools.ts # v1.5: agent register/claim/handoff/lock
│ │ ├── planning-tools.ts # v2.0: plan from prompt/apply/templates
│ │ ├── prediction-tools.ts # v2.0: predict/velocity/burndown
│ │ ├── risk-tools.ts # v2.0: risk assess/report/classify
│ │ ├── sync-tools.ts # v2.0: GitHub/Linear/Jira sync
│ │ ├── pr-tools.ts # v2.0: PR create/status/auto-flow
│ │ ├── approval-tools.ts # v3.0: request/list/resolve
│ │ ├── code-review-tools.ts # v3.0: review run/findings/respond
│ │ ├── qa-tools.ts # v3.0: qa run/report/rerun/coverage
│ │ ├── devops-tools.ts # v3.0: deploy check/run/status
│ │ ├── tech-lead-tools.ts # v3.0: distribute/status/reassign/pipeline
│ │ └── team-tools.ts # v3.0: add/list/assign/workload/suggest
│ ├── storage/ # Atomic writes, file-per-task
│ ├── models/ # Task, Board, Session, Sprint, Config + 8 new models
│ └── utils/ # nanoid, git, project analyzer, file-lock, external-sync
├── vscode-extension/ # VSCode sidebar board view
├── scripts/ # Hook scripts (session, files, commits)
├── commands/ # Slash command definitions
├── skills/ # Smart auto-tracking skill
└── install.sh # One-command installer
Data stored in your project:
.kanban/
├── config.json # Active project + session
├── boards/{id}.json # Board columns (task IDs)
├── tasks/{id}.json # One file per task (with context)
├── sprints/{id}.json # Sprint definitions
├── archive/{id}.json # Completed old tasks
├── sessions/{id}.json # Session logs (gitignored)
├── agents/{id}.json # v1.5: Agent registrations
├── handoffs/{id}.json # v1.5: Handoff contexts
├── locks/{hash}.lock.json # v1.5: File locks
├── history/{id}.json # v2.0: Completion records
├── velocity/{id}.json # v2.0: Velocity snapshots
├── approvals/{id}.json # v3.0: Approval requests
├── reviews/{id}.json # v3.0: Code reviews
├── qa/{id}.json # v3.0: QA results
├── deployments/{id}.json # v3.0: Deployments
└── team/{id}.json # v3.0: Team members
MCP Tools (94)
| Tool | Purpose |
|---|
| Init | |
auto_init | Initialize board + session (idempotent) |
board_summary | One-line status for context injection |
| Smart Tasks | |
task_smart_create | Create task with auto-analysis, tags, priority |
task_create | Create task with title, priority, tags |
task_update | Update task fields |
task_get | Get task by ID or fuzzy name |
task_list | List tasks, filter by status |
task_move | Move between TODO/DOING/DONE |
task_delete | Archive or delete task |
task_prioritize | Change priority + auto-sort column |
task_time | Time tracking report per task or all |
task_analyze | Deep project analysis for a task |
| Context & Agents | |
task_context_save | Save context: files, decisions, remaining work |
task_context_load | Load context when resuming a task |
task_agent_create | Generate .claude/agents/ file for a task |
task_agent_delete | Clean up agent file when done |
task_review | Pre-close: TODOs, tests, changes, type check |
| Dependencies | |
task_depend | Add/remove dependency between tasks |
task_blockers | Show dependency graph for task or all |
critical_path | Find the longest dependency chain (bottleneck) |
| Subtasks | |
task_split |
What install.sh does
- Adds SoloBoard MCP server to
.mcp.json
- Creates hooks in
.claude/settings.json (file tracking, commit linking)
- Injects agent instructions into
CLAUDE.md
- Initializes
.kanban/ directory
- Adds
.kanban/sessions/ to .gitignore
All reversible. Delete these files to uninstall.
Dependencies
Only 3 runtime dependencies:
@modelcontextprotocol/sdk — MCP protocolzod — schema validationnanoid — ID generation
License
MIT
@egorfedorov