by jpicklyk
A light touch MCP task orchestration server for AI agents. Persistent work tracking and context storage across sessions and agents. Defines planning floors through composable notes with optional gating transitions. Coordinates multi-agent execution without prescribing how agents do their work.
# Add to your Claude Code skills
git clone https://github.com/jpicklyk/task-orchestratorStop losing context. Start building faster.
An orchestration framework for AI coding assistants that solves context pollution and token exhaustion — enabling your AI to work on complex projects without running out of memory.
AI agents exhaust their effective context windows on long-running tasks — conversation history, tool outputs, and code examples accumulate until the model loses focus and fails. Task Orchestrator implements industry-recommended patterns to prevent this: persistent external memory, summary-based phase handoffs, and sub-agent isolation with clean contexts. Work scales to 50+ tasks with up to 90% token reduction and zero time rebuilding context after session restarts.
No comments yet. Be the first to share your thoughts!
queue → work → review → terminal with named triggers and automatic dependency enforcementPrerequisite: Docker must be installed and running.
docker pull ghcr.io/jpicklyk/task-orchestrator:latest
This is a one-time step — Docker caches the image locally. Pulling first ensures your MCP client connects instantly rather than waiting silently on first launch.
Choose the option that matches your setup:
Register the server once from your terminal:
claude mcp add-json mcp-task-orchestrator '{
"command": "docker",
"args": [
"run", "--rm", "-i",
"-v", "mcp-task-data:/app/data",
"ghcr.io/jpicklyk/task-orchestrator:latest"
]
}'
R...