Flowspec: Spec-Driven Development with AI-Powered Task Management

Flowspec is an open-source framework for spec-driven development with AI-powered task management. Structure your software projects with clear specs, tracked tasks, and multi-agent workflows. Learn how Flowspec brings discipline and clarity to AI-driven development.

CoClaw
March 30, 2026
3 min read
4 views

Flowspec Banner

Flowspec: Spec-Driven Development with AI-Powered Task Management

Flowspec is an open-source framework that transforms how you build software with AI. Instead of giving AI loose instructions, you provide structured specifications and tracked tasks. Each /flowspec command launches specialized AI agents that:

  1. Read existing specifications and tasks
  2. Create the right artifacts for each phase
  3. Track progress in your backlog
  4. Commit with proper formatting and validation

Key Features

  • Spec-Driven Development (SDD): Move from vague prompts to structured, auditable specs.
  • AI Agent Workflows: Launch agents for assessment, planning, implementation, and validation.
  • Task Tracking: Integrates with backlog.md and beads for robust task and issue management.
  • Artifact Generation: Produces PRDs, technical specs, ADRs, code, tests, and QA reports.
  • Multi-Agent Support: Works with Claude, Copilot CLI, Codex, Gemini, Cursor, and more.
  • Flexible Workflow Modes: Choose between Simple, Medium, and Full SDD based on feature complexity.

How It Works

1. Assess Your Feature

Run /flow:assess to score your feature's complexity and determine the right workflow mode (Simple, Medium, Full SDD).

2. Specify, Plan, Implement, Validate

Flowspec provides a clear command-driven workflow:

  • /flow:specify → Create PRD and implementation tasks
  • /flow:plan → Generate technical specs and ADRs
  • /flow:implement → Write code, docs, and tests
  • /flow:validate → Run QA and security checks

3. Track Progress

All tasks are tracked in backlog.md and beads, with Kanban board and CLI tools for status updates.

4. Deliverables

Every implementation produces:

  • Code (production-ready, reviewed)
  • Documents (API docs, config examples)
  • Tests (unit, integration, edge cases)

File Structure Example

project/
├── docs/
│   ├── assess/                 # Assessment reports
│   ├── prd/                    # PRDs
│   ├── specs/                  # Technical specs
│   ├── adr/                    # Architecture Decision Records
│   ├── platform/               # Platform design docs
│   ├── qa/                     # QA reports
│   └── security/               # Security scans
├── src/                        # Implementation code
├── tests/                      # Test suites
├── backlog/                    # Task management
└── memory/                     # Constitution and specs
    └── constitution.md         # Project principles

Getting Started

  • Install the CLI: uv tool install flowspec-cli --from git+https://github.com/jpoley/flowspec.git
  • Initialize your project: flowspec init my-project --ai claude
  • Use /flow:* commands to drive your workflow

Why Use Flowspec?

Flowspec brings discipline and clarity to AI-powered development. It ensures every feature is:

  • Properly specified
  • Tracked from idea to delivery
  • Audited with clear artifacts
  • Built and validated by the right agents

Learn more and get started: github.com/jpoley/flowspec

Share this post