Skip to content
← Tools / Skills
Skill · .md packagekikmak42/10 trustMetrics-checked
Basic Agent Skills (kikmak42) logo

Basic Agent Skills (kikmak42)

Cross-platform utility skills that patch common LLM blind spots

Visit official site →

About

Eighteen small skills covering things language models reliably get wrong: today's date, random numbers, basic math, base64, UUIDs, unit conversion, hashing, and shell execution, each shipping both PowerShell and Bash scripts with OS detection built in. Guardrails like input validation, secret redaction, and a shell command blocklist come standard, and a pytest eval suite checks the scripts directly. The design is OS-agnostic and works through native skill folders, an MCP server, or adapters for LangChain, CrewAI, AutoGen and others — though the repo is brand new and has a way to go before maturity.

What it can do

  • Timezone-aware dates
  • Cryptographically seeded random numbers
  • Arithmetic without rounding errors
  • Base64 encode/decode
  • UUID and token generation
  • Unit conversion
  • Hash and checksum computation
  • Shell execution with blocklist validation
  • File operations with secret redaction
  • pytest eval suite

Setup

  1. Clone the repository:

git clone https://github.com/kikmak42/basic-agent-skills.git

  1. Copy the skills/ folder into your agent's skills directory (e.g. .agents/skills/ for Antigravity).

Or expose all skills over MCP:

pip install -r integrations/mcp/requirements.txt

python integrations/mcp/server.py

  1. Point your MCP client at the server using integrations/mcp/claude_desktop_config.example.json as a reference.

Prerequisites

Python 3 and the target agent; PowerShell or Bash depending on platform.

Use cases

getting timezone-aware dates without hallucinatingevaluating arithmetic with a real calculatorgenerating UUIDs and secure passwordsadding utility skills to any agent framework via MCP

Works with

Claude CodeCursorGitHub CopilotWindsurfAny MCP client

Trust score — 4/10

/10 trustMetrics-checked

Scored 4/10 (metrics tier, metrics-v1). Evidence: 0 GitHub stars, 7 commits, created 2026-10-02; very new project.; Verified: 18 skills with OS-agnostic ps1/sh scripts, guardrails, pytest evals, MCP server integration.; MIT license confirmed on GitHub license field.. License: MIT License, verified via GitHub license field..

★ 0 GitHub stars

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
1.0/10
Maintenance · recency, releases, issue load
3.0/10
Openness · license + docs
8.0/10
Manual review · sanity read + red-flag check
5.0/10
  • · 0 GitHub stars, 7 commits, created 2026-10-02; very new project.
  • · Verified: 18 skills with OS-agnostic ps1/sh scripts, guardrails, pytest evals, MCP server integration.
  • · MIT license confirmed on GitHub license field.

Sources

Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.