andrew54068

Analyzes project and generates Claude Code permissions for detected tech stacks.

1 skill · 1 command

Turn vague task ideas into well-structured Ralph Loop prompts. Generates completion criteria, suggests iteration limits, previews the full prompt, then launches the ralph-loop with safety nets. Requires the ralph-loop plugin.

1 skill · 1 command

Manage per-project skills, plugins, and MCP servers via a bullpen + symlink pattern. Vendor-neutral path (~/.agents/) works for both Claude Code and Codex.

1 skill · 1 command

Selects the best browser MCP tool for a given task based on requirements like cross-browser support, debugging depth, performance analysis, and login state needs.

1 skill

Prepare a private codebase for safe open-source release with secret scanning, license guidance, and standard repository documentation.

1 skill

Comprehensive 6-agent parallel security scanner for Claude Code. Detects malicious code, credential leaks, data exfiltration, supply chain risks, and prompt injection attacks.

1 command

Preview the images and markdown files of a conversation in Claude Code: hover cards, pasted-image thumbnails above the prompt, and a side pane that draws markdown as a preview does. Pictures show over ssh, and over mosh in block characters.