MCP App StoreMCP App Store
from snip-skill10
Drastically reduce LLM token usage by filtering verbose CLI output without losing critical debugging information.
This skill provides comprehensive guidance on using snip, a CLI proxy that sits between an AI coding assistant and the shell. It optimizes token consumption by condensing verbose command outputs (e.g., transforming long test results into simple pass/fail summaries), potentially saving 60-90% of tokens per command.
Activate this skill when setting up snip, configuring token-saving filters, debugging cases where the agent is missing information due to aggressive filtering, or implementing the 'Escape Protocol' to bypass filters for critical commands.
Claude Code, Cursor, Copilot, Gemini CLI, and OpenCode.
This skill has not been reviewed by our automated audit pipeline yet.
Caveman Compress
Compress natural-language memory files into a compact 'caveman' format to save LLM tokens while preserving code, links, and structure.
Cost Tracker
Monitor agent session costs, set budget alerts, and get actionable token-spend optimizations to keep multi-session workflows within budget.
Copilot Instructions Blueprint Generator
Generates a technology-agnostic blueprint for creating copilot-instructions.md files that align Copilot output with a project's exact architecture, versions, an
Minimalist Code Writer
A high-signal coding standard focused on minimizing noise, eliminating redundant comments, and removing narrative output.
SDD Onboard
Interactive onboarding that guides a user through a full SDD (spec-driven development) cycle on their real codebase, producing proposal, specs, design, tasks, a
API Documentation Generator
Generates clear, consumable API docs (OpenAPI, endpoint guides, examples) from source code to speed API documentation and SDK examples.
Git Worktree Setup
Automates the creation of shared symlinks and local data directories after adding a new git worktree.
Setup Skill Discovery Wizard
Generates or updates a project-local .dh/skill_discovery.yaml by scanning a repo, inferring stack signals, and suggesting relevant skills — supports both intera