Current
gmickel Claude Marketplace
A GitHub-hosted plugin marketplace extending Claude Code with autonomous workflow patterns, multi-model review gates, and receipt-based gating for reliable AI coding execution.
Signal
gmickel Claude Marketplace · 2026-03-22
GitHub repository gmickel/gmickel-claude-marketplace published 2026-03-22. Provides a collection of plugins for Claude Code focused on reliable AI coding workflows. Key components include Flow-Next for plan-first execution, Ralph autonomous mode for overnight coding with fresh context, and receipt-based gating to prevent drift. Compatible with Factory Droid runtime.
Context
Claude Code is an autonomous coding interface that operates via command-line and IDE integration. This marketplace extends the base capabilities by introducing structured workflow plugins rather than ad-hoc prompting. The infrastructure shifts from single-turn interaction to multi-step orchestration with built-in review gates and state management.
Relevance
Standard coding agents often suffer from context drift and lack of execution verification. This entry addresses reliability through receipt-based gating and cross-model review gates (RepoPrompt/Codex). It represents a shift toward treating AI coding agents as production-grade infrastructure requiring explicit workflow constraints and audit trails.
Current State
The primary plugin set includes Flow-Next (v0.26.1) which enforces plan-first workflows before execution. Ralph mode enables autonomous overnight operation with automatic task blocking if stuck. Cross-platform reviews allow macOS (RepoPrompt) or generic OS (Codex CLI) validation of generated code. The system supports Epic-completion review gates to catch requirement gaps before task closure.
Open Questions
- Long-term maintenance of the plugin ecosystem independent of the author.
- Security implications of overnight autonomous execution modes with elevated permissions.
- Comparison of drift prevention mechanisms against other agent frameworks like OpenClaw or AutoGen.
- Integration depth with existing CI/CD pipelines beyond local repository validation.
Connections
- anthropic-cybersecurity-skills: Compatible with Claude Code for skill integration, indicating shared runtime constraints and API surface.
- multi-agent-coding-orchestration: Parallel approach to coordinating specialized coding agents, though this entry focuses on single-agent workflow depth rather than swarm coordination.
- openclaw: Comparative agent framework for inspectable workflows, offering a contrast between marketplace plugin architecture and monolithic framework design.