gstack-coder

Orchestrates Claude Code, gstack, and mini-swe-agent for multi-tool coding workflows.

Updated Sep 2, 2026
One-click install
npx skills add https://github.com/nbiish/tts-mcp --skill gstack-coder-nbiish
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gstack-coder
Source: https://github.com/nbiish/tts-mcp/tree/main/.agents/skills/gstack-coder
Command: npx skills add https://github.com/nbiish/tts-mcp --skill gstack-coder-nbiish

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating multiple AI coding tools (Claude Code, gstack skills, mini-swe-agent) across subscriptions and quotas is manual and error-prone; this Skill standardizes how to install, configure, and dispatch each tool for security audits, code reviews, QA testing, and feature builds. ## Core Features & Use Cases - GStack Slash Commands: Run /cso for security audits, /review for code reviews, /qa for URL testing, /autoplan for planning, and /ship for build-and-deploy pipelines inside Claude Code. - Multi-Tool Dispatch: Execute sequential or parallel coding tasks across gemini, claude, mini-live, and other agents using git worktrees for isolation. - Secret & Provider Injection: Load API keys via secrets-load before invoking tools, with PQC-compliant secrets handling. - Use Case: You need a full security review of a repository — load gstack and run /cso, then dispatch mini-live to fix the findings with mini-live --task "Fix the race condition" --yolo. ## Quick Start Ask the agent to load gstack and run /cso to perform a full security audit of the current repository.

Frequently Asked Questions about gstack-coder

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a security audit with Claude Code and gstack?▼

Install gstack into ~/.claude/skills/gstack, then inside a Claude Code session load gstack and run /cso. It performs a full codebase security review covering injection vectors, secret leaks, and dependency vulnerabilities.

How to dispatch coding tasks to multiple AI agents in parallel?▼

Create separate git worktrees per task, then launch each agent (gemini, claude, mini-live) in its own worktree with non-interactive flags like -y or --yolo. Run secrets-load first so API keys are available, then merge the worktrees after review.

What is the difference between mini and mini-live in mini-swe-agent?▼

mini-live is a shell function that runs mini with the Live-SWE-agent config at ~/.config/mini-swe-agent/live-swe-agent.yaml. This skill uses mini-live for all orchestrated SWE work; bare mini is reserved for upstream default behavior only.

Can I plan a feature without implementing it using gstack?▼

Yes. Load gstack, run /office-hours for architecture discussion, then /autoplan to generate a detailed implementation plan. Save the plan and skip implementation — no code changes are made.

Why must secrets-load run before invoking claude or mini-live?▼

API keys and provider endpoints are stored in a PQC secrets bundle and are not in the shell environment by default. Running secrets-load exports them into memory so claude and mini-live can authenticate with their providers.