agentskill-kit CLI
The `askit` command line toolkit for agent skills: lint, security scan, tests, forging new ones and export to any agent.
- 50 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Skill tests run on your machine, `askit mine` reads shell history and `askit install` writes into agent directories.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/agentskill-kitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Skill tests run on your machine, `askit mine` reads shell history and `askit install` writes into agent directories.
This is third-party code. Review the repository files before installing.
What it does
agentskill-kit checks a skill before you trust it, creates new ones and installs the same skill into every agent you use. It installs from PyPI as `agentskill-toolkit` and adds the `askit` command. Without keys, `askit lint` checks SKILL.md structure, `askit scan` looks for prompt injection, hidden unicode, secrets and dangerous commands, `askit test` runs `tests/cases.json` and `tests/test_*.py` in an isolated folder and `askit score` combines checks into one score. `askit forge` writes a skill from a task description on any model, including a free local one through Ollama, verifies it and refines it until checks pass. `askit mine` finds repeated routines in shell history, `askit export` converts a skill for Claude Code, Cursor, Windsurf, Copilot, Codex and Gemini CLI, and the registry offers search and installation of others skills with checksum verification.
Who it is for. Authors and teams who write and select agent skills and want to check them automatically and install them into several agents at once.
Good fit when
- A skill from a third-party repository must be checked for injection and hidden unicode before installing
- CI needs a gate: skill structure, security and tests must pass
- A task was described in words and has to become a skill with tests on a free model
- A skill is written for Claude Code and Cursor and Gemini CLI need it too
- Shell history shows a repeated routine worth turning into a skill
Not a fit when
- You need a full sandbox: tests run in an isolated folder but not in a container
- A skill is plain text and security checks must catch sophisticated attacks: the scanner looks for common signs
- Python older than 3.9 is installed: the package needs 3.9 or newer
Example request
Lint and scan this skill, run its tests and export it to Claude Code and Cursor if the score passes the gate.Limitations
The scanner looks for common signs: prompt injection, hidden unicode, secrets and dangerous commands. It does not replace a manual review of a third-party skill. Skill tests run in a clean folder but not a container, so skill code executes on your machine. `askit mine` reads shell history, which may hold private commands and paths. `askit forge` sends the task description to the chosen model, and with a cloud provider the key is set through a variable such as `GROQ_API_KEY`. The package is named `agentskill-toolkit` because the short name was taken, and the `askit` and `agentskill` commands install together.
How to disable. Remove the package with `pip uninstall agentskill-toolkit`. Skills installed through `askit install` or `askit export` can be removed with `askit uninstall` or manually from the agent directory.
FAQ
How do I install agentskill-kit?
Run `pip install agentskill-toolkit`. The package adds the `askit` and `agentskill` commands. Python 3.9 or newer is required.
Do I need API keys?
Not for `lint`, `scan`, `test` and `score`. `forge` needs a model: local Ollama or LM Studio without a key, free tiers of Groq and Gemini, OpenRouter or any OpenAI-compatible server.
What does `askit scan` catch?
Prompt injection, hidden unicode, secrets and dangerous commands. The failure threshold is set with `--fail-on` using high, medium, low or never.
How do I install a skill into all agents?
With `askit export skills/<name> --target all`. You can also name one agent such as claude or cursor.
How are registry skills verified?
Downloads are verified against checksums, then the skill is linted and scanned before installation.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI skill manager for AI agents: install from repositories, translate between agent formats and deploy into their configs
A CLI that compiles a SKILL.md file into a standalone Python agent with tools, knowledge-base retrieval and its own runner.
A command-line skill manager: installs skills from a marketplace, GitHub, private Git and npm into 45 agents' folders and scans them first
A skill that helps an agent author other skills to the agentskills.io spec: structure, metadata, progressive disclosure and validation
A CLI and Python SDK to search, install, create and evaluate reusable agent skills from a shared library
A CLI and manager skill: one skills library distributed to 70+ agents, with management done from the agent conversation
README in short
The README lists toolkit capabilities, installation, a keyless quick start, a table of commands and free model providers, plus how `forge` works and the skill registry.