Skip to content

Speckit reimagined: Claude Code plugins with agent skills and multi-agent workflows. #1425

@deepeshBodh

Description

@deepeshBodh

Speckit reimagined: Claude Code plugins with agent skills that turn specs into autonomous, multi-agent workflows

→ https://github.com/deepeshBodh/human-in-loop


We took inspiration from speckit and rebuilt it from the ground up as a Claude Code plugin marketplace. The result is humaninloop—a spec-driven development workflow where specialized agents collaborate to take you from vague feature request to production code.

What's different

Multi-agent architecture Requirements Analyst, Devil's Advocate, Plan Architect, Task Architect—each owns their phase
Agent skills Auto-triggered capabilities like authoring-requirements, entity-modeling, vertical-tdd that activate based on context
Validation loops Agents challenge each other before code gets written
Full pipeline /setup → /specify → /plan → /tasks → /implement

Not a port. Not a migration. A fundamental rearchitecture built for Claude Code's plugin system.

→ https://github.com/deepeshBodh/human-in-loop

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions