All notable changes to Mission Control will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
1.4.0 - 2026-03-03
- Multi-Agent Workflow Pipeline — Full task lifecycle now supports staged orchestration:
planning → inbox → assigned → in_progress → testing → review → verification → done. - Core Agent Bootstrap — New workspaces can auto-bootstrap a 4-agent core team: Builder (🛠️), Tester (🧪), Reviewer (🔍), and Learner (📚).
- Workflow Engine Coordination — Added queue-aware review draining (
drainQueue()), automatic role-based stage handoffs, and fail-loopback routing. - Learner Knowledge Loop — Learner notifications on stage transitions plus knowledge injection into future dispatch messages.
- New API Routes
POST /api/tasks/[id]/failGET /api/tasks/[id]/rolesPOST /api/workspaces/[id]/knowledgeGET /api/workspaces/[id]/workflows
- Strict template defaults — Strict workflow is now default, with review as queue stage and verification owned by the
reviewerrole. - Workspace initialization — New workspaces can clone workflow templates and bootstrap core agents automatically.
- Project branding/docs — Updated project branding to Autensa (formerly Mission Control) and added explicit privacy-first statement in docs.
- Role mismatch — Fixed strict template verification role (
verifier→reviewer). - Review queue bypass — Fixed auto-advance behavior that could skip proper review queue flow.
- Dispatch status transition — Fixed dispatch route using hardcoded
done; now uses computed next workflow status. - Assigned-status resolution — Fixed mapping so
assignedresolves to builder stage dispatch correctly. - Task template assignment — Fixed task creation path so default workflow template is attached automatically.
- Learner role assignment — Fixed missing
task_roleslearner assignment so the learner receives transition events.
- Migration 013: Fresh Start — Resets runtime task/agent/event data, sets Strict as default workflow template, and bootstraps core agents for the default workspace.
1.3.0 - 2026-03-02
- Agent Activity Dashboard — Dedicated page for monitoring agent work with mobile card layout. (#48 — thanks @pkgaiassistant-droid!)
- Remote Model Discovery — Discover AI models from OpenClaw Gateway via
MODEL_DISCOVERY=trueenv var. (#43 — thanks @davetha!) - Proxy Troubleshooting — Added docs for users behind HTTP proxies experiencing 502 errors on agent callbacks.
- Force-Dynamic API Routes — All API routes now use
force-dynamicto prevent stale cached responses. (#43) - Null Agent Assignment —
assigned_agent_idcan now be null in task creation schema. (#38 — thanks @JamesCao2048!) - Dispatch Spec Forwarding — Planning spec and agent instructions now forwarded in dispatch messages. (#51)
- Dispatch Failure Recovery — Tasks stuck in
pending_dispatchauto-reset to planning status. (#52)
1.2.0 - 2026-02-19
- Gateway Agent Discovery — Import existing agents from your OpenClaw Gateway into Mission Control. New "Import from Gateway" button in the agent sidebar opens a discovery modal that lists all Gateway agents, shows which are already imported, and lets you bulk-import with one click. Imported agents display a
GWbadge for provenance tracking. (#22 — thanks @markphelps!) - Docker Support — Production-ready multi-stage Dockerfile, docker-compose.yml with persistent volumes, and
.dockerignore. Runs as non-root, usesdumb-initfor signal handling, includes health checks. (#21 — thanks @muneale!) - Agent Protocol Conventions — Added
PROGRESS_UPDATEandBLOCKEDmessage formats to the Agent Protocol docs to prevent agent stalling. (#24 — thanks @nice-and-precise!)
- Planning Flow Improvements — Refactored polling to prevent stale state issues, fixed "Other" free-text option (case mismatch bug), made
due_datenullable, increased planning timeout to 90s for larger models, auto-start polling on page load. (#26 — thanks @JamesTsetsekas!) - WebSocket RPC Deduplication Bug — The event deduplication cache was silently dropping repeated RPC responses with the same payload hash, causing request timeouts. RPC responses now bypass dedup entirely.
- Next.js Response Caching — Dynamic API routes that query live state (e.g., agent discovery) now use
force-dynamicto prevent stale cached responses.
1.1.0 - 2026-02-16
- API Authentication Middleware — Bearer token authentication for all API routes. Set
MC_API_TOKENin.env.localto enable. Same-origin browser requests are automatically allowed. - Webhook HMAC-SHA256 Validation — Agent completion webhooks now require a valid
X-Webhook-Signatureheader. SetWEBHOOK_SECRETin.env.localto enable. - Path Traversal Protection — File download endpoint now uses
realpathSyncto resolve symlinks and validate all paths are within the allowed directory. - Error Message Sanitization — API error responses no longer leak internal details (stack traces, file paths) in production.
- Security Headers — Added
X-Frame-Options,X-Content-Type-Options,Referrer-Policy, andPermissions-Policyheaders via Next.js config. - Input Validation (Zod) — Request payloads for tasks, agents, and workspaces are validated with Zod schemas before processing.
- Repository Audit — Purged sensitive files from git history, updated
.gitignoreto block database files and backups.
- Ed25519 Device Identity — Gateway pairing now uses Ed25519 key-based device identity for secure handshakes.
- ARIA Hook — Real-time agent tracking bridge between ARIA and Mission Control (
scripts/aria-mc-hook.sh). - Planning Poll Endpoint — New
POST /api/tasks/[id]/planning/pollfor long-poll planning updates. - Retry Dispatch — New
POST /api/tasks/[id]/planning/retry-dispatchto retry failed task dispatches. - Auto-Dispatch Module —
src/lib/auto-dispatch.tsfor automatic task assignment after planning. - Planning Utilities —
src/lib/planning-utils.tswith shared planning logic. - MC Bridge Scripts — Python and shell bridge scripts for external integrations.
- Node.js v25 Support — Updated
better-sqlite3to v12.6.2 for Node v25 compatibility. - Default Port — Mission Control now defaults to port 4000 (previously 3000).
- Improved Planning Tab — Enhanced UI with better question rendering, progress tracking, and error handling.
- Agent Sidebar Improvements — Better status display, model selection, and agent management.
- Activity Log Overhaul — Cleaner timeline UI with better type icons and formatting.
- Live Feed Improvements — Better real-time event display with filtering options.
- Same-origin browser requests — Auth middleware no longer blocks the UI's own API calls.
1.0.1 - 2026-02-04
- Clickable Deliverables - URL deliverables now have clickable titles and paths that open in new tabs
- Improved visual feedback on deliverable links (hover states, external link icons)
1.0.0 - 2026-02-04
This is the first stable, tested, and working release of Mission Control.
-
Task Management
- Create, edit, and delete tasks
- Drag-and-drop Kanban board with 7 status columns
- Task priority levels (low, normal, high, urgent)
- Due date support
-
AI Planning Mode
- Interactive Q&A planning flow with AI
- Multiple choice questions with "Other" option for custom answers
- Automatic spec generation from planning answers
- Planning session persistence (resume interrupted planning)
-
Agent System
- Automatic agent creation based on task requirements
- Agent avatars with emoji support
- Agent status tracking (standby, working, idle)
- Custom SOUL.md personality for each agent
-
Task Dispatch
- Automatic dispatch after planning completes
- Task instructions sent to agent with full context
- Project directory creation for deliverables
- Activity logging and deliverable tracking
-
OpenClaw Integration
- WebSocket connection to OpenClaw Gateway
- Session management for planning and agent sessions
- Chat history synchronization
- Multi-machine support (local and remote gateways)
-
Dashboard UI
- Clean, dark-themed interface
- Real-time task updates
- Event feed showing system activity
- Agent status panel
- Responsive design
-
API Endpoints
- Full REST API for tasks, agents, and events
- File upload endpoint for deliverables
- OpenClaw proxy endpoints for session management
- Activity and deliverable tracking endpoints
- Built with Next.js 14 (App Router)
- SQLite database with automatic migrations
- Tailwind CSS for styling
- TypeScript throughout
- WebSocket client for OpenClaw communication
0.1.0 - 2026-02-03
- Initial project setup
- Basic task CRUD
- Kanban board prototype
- OpenClaw connection proof of concept
- Multiple workspaces
- Webhook integrations
- API authentication & security hardening
- Team collaboration
- Task dependencies
- Agent performance metrics
- Mobile-responsive improvements
- Dark/light theme toggle