TaskWing

Give your AI tools a brain.

Memory, planning, task execution, and project intelligence — the control plane for AI-native development.

Website · Tutorial · Vision · Install

---

The Problem

Your AI tools start every session from zero. They don't know your stack, your patterns, or why you chose PostgreSQL over MongoDB. You re-explain the same context hundreds of times.

TaskWing fixes this. One command extracts your architecture into a local database. Every AI session after that just knows.

Why TaskWing?

Your AI assistant reads the same files every session. TaskWing remembers so it doesn't have to.

Without TaskWing              With TaskWing
─────────────────             ─────────────
8–12 file reads               1 MCP query
~25,000 tokens                ~1,500 tokens
2–3 minutes                   42 seconds
Zero persistent context       170+ knowledge nodes

Real session, real numbers — asked "What are the bottlenecks in our engineering process?":

• Without TaskWing: 8 Glob/Grep searches, 12 file reads, 25,000 tokens, 3 minutes
• With TaskWing MCP: 1 query, 1,500 tokens, 42 seconds — synthesized answer with code references

That's 90% fewer tokens and 75% faster time-to-answer.

What It Does

Capability	Description
🧠 Memory	Extracts and persists architectural decisions, patterns, and constraints
📋 Planning	Turns a goal into an executable plan with decomposed tasks
⚡ Task Execution	AI-driven task lifecycle — next, start, complete, verify
🔍 Code Intelligence	Symbol search, call graphs, impact analysis, simplification
🐛 Debugging	AI-powered root cause analysis with systematic diagnosis
🔌 MCP Integration	Exposes everything to Claude, Cursor, Copilot, Gemini via MCP

Install

# Homebrew (recommended)
brew install josephgoksu/tap/taskwing

# or curl
curl -fsSL https://taskwing.app/install.sh | sh

No signup. No account. Works offline. Everything stays local in SQLite.

Supported Models

Works With

Brand names and logos are trademarks of their respective owners; usage here indicates compatibility, not endorsement.

Quick Start

# 1. Extract your architecture
cd your-project
taskwing bootstrap
# → 22 decisions, 12 patterns, 9 constraints extracted

# 2. Set a goal and generate a plan
taskwing goal "Add Stripe billing"
# → Plan decomposed into 5 executable tasks

# 3. Execute with your AI assistant
/taskwing:next       # Get next task with full context
# ...work...
/taskwing:done       # Mark complete, advance to next

MCP Tools

Tool	Description
ask	Search project knowledge (decisions, patterns, constraints)
task	Unified task lifecycle (next, current, start, complete)
plan	Plan management (clarify, decompose, expand, generate, finalize, audit)
code	Code intelligence (find, search, explain, callers, impact, simplify)
debug	Diagnose issues systematically with AI-powered analysis
remember	Store knowledge in project memory

MCP Setup

Add to your AI tool's MCP config:

{
  "mcpServers": {
    "taskwing": {
      "command": "taskwing",
      "args": ["mcp"]
    }
  }
}

Slash Commands

Once connected, use these slash commands from your AI assistant:

Command	When to use
/taskwing:ask	Search project knowledge (decisions, patterns, constraints)
/taskwing:remember	Persist a decision, pattern, or insight to project memory
/taskwing:next	Start the next approved task with full context
/taskwing:done	Complete the current task after verification
/taskwing:status	Check current task progress and acceptance criteria
/taskwing:plan	Clarify a goal and build an approved execution plan
/taskwing:debug	Root-cause-first debugging before proposing fixes
/taskwing:explain	Deep explanation of a code symbol and its call graph
/taskwing:simplify	Simplify code while preserving behavior

Core Commands

• taskwing bootstrap
• taskwing goal "<goal>"
• taskwing ask "<query>"
• taskwing task
• taskwing plan status
• taskwing slash
• taskwing mcp
• taskwing doctor
• taskwing config
• taskwing start

Autonomous Task Execution (Hooks)

TaskWing integrates with Claude Code's hook system for autonomous plan execution:

taskwing hook session-init      # Initialize session tracking
taskwing hook continue-check    # Check if should continue to next task
taskwing hook session-end       # Cleanup session
taskwing hook status            # View current session state

Circuit breakers prevent runaway execution:

• --max-tasks=5 — Stop after N tasks for human review
• --max-minutes=30 — Stop after N minutes

AWS Bedrock Setup

TaskWing supports Bedrock as a first-class provider:

llm:
  provider: bedrock
  model: anthropic.claude-sonnet-4-5-20250929-v1:0
  bedrock:
    region: us-east-1
  apiKeys:
    bedrock: ${BEDROCK_API_KEY}

Recommended Bedrock models

Model	Use case
anthropic.claude-opus-4-6-v1	Highest quality reasoning
anthropic.claude-sonnet-4-5-20250929-v1:0	Best default balance
amazon.nova-premier-v1:0	AWS flagship Nova
amazon.nova-pro-v1:0	Strong balance
meta.llama4-maverick-17b-instruct-v1:0	Open-weight general model

Or configure interactively: taskwing config

Documentation

• Getting Started
• Product Vision
• Architecture
• Workflow Contract v1
• Workflow Pack

License

MIT