Releases: egorfedorov/Soloboard
v3.0.1
v3.0.0 — Autonomous Development Orchestrator
What's New — 22 new tools (72→94)
SoloBoard evolves from a task tracker into an autonomous development orchestrator. Full pipeline: coding → review → QA → deployment, all managed locally.
Tech Lead Tools
lead_distribute— analyze TODO tasks, distribute to agents by dependencies, complexity, and skillslead_status— dashboard: all agents, tasks, pipeline progress, recent reviews/QA/deployslead_reassign— reassign a task with full handoff contextlead_pipeline— view and manage the full pipeline: coding → review → QA → deploy
Automated Code Review
review_run— analyze changed files for TODOs, type errors, security issues, console statementsreview_findings— view detailed findings filtered by severityreview_respond— respond to findings: fixed / wont_fix / acknowledged
QA & Testing
qa_run— run tests, parse results, auto-create bug tasks for failuresqa_report— view QA results with pass/fail/skip countsqa_rerun— re-run tests after fixes, compare with previous runqa_coverage— check test file coverage for changed files
DevOps & Deployment
deploy_check— readiness check: all tasks done, reviewed, QA passeddeploy_run— execute deployment command (approval required for production)deploy_status— deployment history and status
Human-in-the-Loop Approvals
approval_request— create approval request for human reviewapproval_list— list pending approval requestsapproval_resolve— approve or reject with reason
Team Management
team_add— add team member with role and skillsteam_list— list members with stats and current assignmentsteam_assign— assign task to a team memberteam_workload— workload distribution view with chartteam_suggest_assignment— auto-suggest best member based on skills and availability
VSCode Extension
New vscode-extension/ package:
- TreeView board in sidebar (TODO/DOING/DONE columns)
- Team panel showing members and their tasks
- FileSystemWatcher for live updates on
.kanban/changes - Refresh command
New Data
.kanban/approvals/— approval request files.kanban/reviews/— code review result files.kanban/qa/— QA test result files.kanban/deployments/— deployment record files.kanban/team/— team member files- Tasks now have
assignedMemberId,reviewStatus,qaStatus,deploymentId - Config supports
teamMode,roles,deployCommand,autoReview,autoQA,autoDeploy
Roles
Built-in roles: tech_lead, code_reviewer, qa_agent, devops_agent, developer
Full changelog: 44 → 94 tools across v1.5, v2.0, and v3.0
v2.0.0 — AI-native Project Manager
What's New — 18 new tools (54→72)
Natural Language Planning
plan_from_prompt— describe what you want to build → structured task breakdown with dependencies and estimatesplan_apply— bulk-create tasks from a plan, set up dependencies, optionally create a sprintplan_templates— pre-built templates for SaaS, API, CLI, and library projects
Predictive Estimates & Velocity
predict_duration— predict task time based on historical similar tasksvelocity_report— tasks/day trends over 7 days, sprint projectionburndown_data— ASCII burndown chart for active sprintrecord_velocity— snapshot daily velocity (also auto-called on task completion)
Risk Assessment
risk_assess— analyze task risk from git hotspots, dependency depth, complexity, and sensitive areasrisk_report— all active tasks ranked by risk score with mitigationscomplexity_classify— auto-classify tasks as trivial/small/medium/large/epic
External Tool Sync
sync_setup— configure GitHub / Linear / Jira credentialssync_push— push a task to external tool as an issuesync_pull— import issues from external tool into SoloBoard taskssync_update— sync status changes bidirectionallysync_status— show sync state for all linked tasks
Pull Request Automation
pr_create— create branch + push + open PR + link to taskpr_status— check PR review/CI/merge statuspr_auto_flow— full flow: branch → commit → push → PR → link to task
Auto-Recording
Task completions are now automatically recorded to history for predictions and velocity tracking.
New Data
.kanban/history/— completion records with time and complexity.kanban/velocity/— daily velocity snapshots- Tasks now have
externalLinks,predictedMinutes,complexity,riskLevel - Config supports
githubSync,linearSync,jiraSync
v1.5.0 — Multi-Agent Orchestration
What's New — 10 new tools (44→54)
Multi-Agent Support
Multiple Claude Code agents can now work on the same project simultaneously.
agent_register— register an agent session for multi-agent workagent_heartbeat— update heartbeat, auto-clean stale agents (>5min inactive)agent_list— list all active agents with their tasks and locked filesagent_claim_task— assign task to agent (fails if already claimed by another)
File Locking
conflict_check— check if files are locked by another agent before editingfile_lock— lock files to prevent concurrent editsfile_unlock— release file locks (specific files or all)
Agent Handoff
agent_handoff— create handoff context with summary, decisions, remaining work + release all locksagent_pickup— accept a handoff and get full context from previous agent
Parallel Planning
parallel_plan— analyze dependency graph and suggest parallelizable task batches
New Data
.kanban/agents/— agent registration files.kanban/handoffs/— handoff context files.kanban/locks/— advisory file lock files- Tasks now have
assignedAgentIdfield - Config has
multiAgentEnabledandmaxParallelAgentssettings
v1.4.0 — Auto-Manager Agent
What's New — 5 new tools (34→44)
Autonomous Project Manager
manager_report— project health score (0-100), velocity, stall alerts, recommendationsstall_detect— find tasks with no recent activitysuggest_next— AI-powered "what to work on next" with weighted scoringauto_reprioritize— smart priority adjustment based on blockers & progressgantt_view— text-based Gantt chart with dependency visualization
How It Works
The manager analyzes your board continuously:
- Health score factors in velocity, stalled tasks, blocked tasks, WIP limits
- Suggestions consider priority, blocking chains, and time in status
- Reprioritization can run in dry-run mode to preview changes
v1.3.0 — Dependencies & Planning
What's New — 12 new tools (22→34)
Task Dependencies
task_depend— add/remove dependency between tasks with circular dependency detectiontask_blockers— show full dependency graphcritical_path— find the longest dependency chain (project bottleneck)
Task Splitting
task_split— break complex tasks into subtaskstask_subtasks— view subtask progress with completion percentage
Sprint Planning
sprint_create— create time-boxed sprints (default 7 days)sprint_add— add tasks to a sprintsprint_close— close sprint, carry over incomplete taskssprint_view— sprint progress with burndown
Daily Standup & Focus
standup— automated standup: done, in-progress, blocked, up nextpomodoro_start— start focus session tied to a taskpomodoro_status— check pomodoro timer
v1.2.0 — Intelligence
What's New
Smart Task Creation
task_smart_create analyzes your project when creating a task:
- Finds related files by searching filenames and code
- Auto-tags based on content (#bug, #feature, #performance, #security, etc.)
- Sets priority from urgency keywords and git hotspots
- Suggests approach: which files to check, related commits, missing tests
Task Context (Memory)
Never lose context when switching tasks:
task_context_save— saves files viewed, decisions made, remaining worktask_context_load— restores full context when resuming a task- Context auto-saved when switching between tasks
Task Agents
Auto-generate .claude/agents/ files for complex tasks:
task_agent_create— creates agent with full task context, key files, constraints- Launch in any Claude Code session with
/agents task_agent_delete— cleans up when task is done
Auto-Review
Pre-close analysis before marking a task done:
task_review— checks for TODOs/FIXMEs, verifies tests exist, counts changes- Type check status (TypeScript projects)
- Remaining work from saved context
- Clear pass/fail checklist
Deep Analysis
task_analyze— project-level analysis: git history, test coverage, code hotspots- Auto-detect tags from keywords and file types
- Priority suggestion based on git churn data
Stats
- 27 MCP tools (up from 20)
- 7 new tools in the Intelligence category
- Zero new dependencies
v1.1.0 — Time Tracking, Dashboard, Export, CLI
SoloBoard v1.1.0
New Features
Time Tracking
- Automatic: time starts when task moves to DOING, stops when it leaves
task_timeshows per-task and total time reports- Time displayed on board_view next to each task
Priority Sorting
- Board columns auto-sorted: high → medium → low
task_prioritizetool to change priority + re-sort
Markdown Export
board_exportgenerates clean markdown with checkboxes, priorities, tagsboard_exportwithincludeDetails: trueadds git info, files, descriptions
Multi-Project Dashboard
dashboardshows all projects with task counts and active tasks- Total time tracked across all projects
Global CLI
npm install -g soloboard
soloboard install # install into current project
soloboard server # start MCP server
soloboard helpStats
- 20 MCP tools (was 16)
- 4 new tools:
board_export,dashboard,task_prioritize,task_time - CLI binary:
soloboard
v1.0.0 — Initial Release
SoloBoard v1.0.0
Invisible personal kanban for Claude Code. You code — board tracks.
What's included
- 16 MCP tools — task CRUD, board management, session tracking, git integration
- Auto-tracking — actionable prompts silently become tasks, questions are ignored
- Git integration — commits and branches auto-linked to active task via hooks
- Fuzzy search — find tasks by partial name, not just ID
- File-per-task storage — each task is a JSON file in
.kanban/, git-friendly - One-command install —
bash install.sh /path/to/project - 3 slash commands —
/soloboard:board,/soloboard:task,/soloboard:project - Landing page —
index.htmlwith full docs
Install
git clone https://github.com/egorfedorov/Soloboard.git
cd Soloboard && npm install && npm run build
bash install.sh /path/to/your/projectThen just cd your-project && claude — the board manages itself.
Tech
- MCP Server (stdio transport)
- TypeScript, 3 deps:
@modelcontextprotocol/sdk,zod,nanoid - Atomic writes, zero in-memory state
- Claude Code hooks for file tracking + commit linking