oil-skill-creator
A skill that creates and reviews other skills, weighing usefulness, trigger boundaries, setup and stability before release.
- 102 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The optional key entry page stores keys in the system credential store on your machine.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/oil-skill-creatorDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The optional key entry page stores keys in the system credential store on your machine.
This is third-party code. Review the repository files before installing.
What it does
oil-skill-creator, a skill for Python 3.10 and later, works in three modes: it creates a new skill, reviews one without editing files and improves an existing one. It first checks whether the skill solves a repeating task better than a plain agent. A review returns a report with P0, P1 and P2 priorities. The bundle has standard-library scripts: `scaffold_skill.py` creates a skeleton, `validate_skill.py` checks structure, links, personal paths and plaintext keys, `score_triggers.py` measures trigger accuracy on positive and negative examples and `package_skill.py` builds a `.skill` package. A separate section covers paid release: the free and paid split, payment and updates.
Who it is for. Skill authors for Claude Code, Codex and other agents who prepare a skill for public release and want to check it before shipping.
Good fit when
- A recurring team task becomes a skill from scratch: skeleton, evaluations and comparison with a plain agent.
- A finished skill fires on the wrong requests, and triggers are tested on positive and negative examples.
- Before publishing on GitHub a skill is reviewed read-only and returns a prioritized report.
- A skill needs an API key, and instead of chat it is entered on a local page with system keychain storage.
Not a fit when
- The task is one-off or a plain agent already handles it reliably: the README advises against packaging that as a skill.
- The target skill's actual task must be done right now: oil-skill-creator does not solve it.
- A human judges design and copy: model self-scoring and scripts do not replace that.
Example request
Review the skill in this folder read-only: a list of issues with P0, P1, P2 priorities and evidence.Limitations
The documentation is in Chinese. Testing ran on macOS, while Windows and Linux await real runs. An independent comparison with a plain agent needs isolated executors such as subagents, and without them only static review and tests are available. The optional key entry page needs Node.js 22.18 or later and a system credential store. An agent allowed to execute code as your user can still reach the stored keys.
How to disable. Remove the skill folder from the agent's skills directory. Delete snapshots, evaluation folders and `.skill` packages created in projects separately. Keys from the entry page sit in the system credential store.
FAQ
Does a review change files?
No. In review mode files stay untouched, no snapshots are made and nothing is packaged. Edits start after a separate request.
What does it need?
Python 3.10 or later for scripts and `npx` for the command install. Keys and setup are not needed for the core flow.
How is skill triggering checked?
With `score_triggers.py` on positive, negative and held-out request sets, not by keywords.
What does the key entry page do?
It takes 1 to 16 keys on a local page and stores them in the system store: macOS Keychain, Windows Credential Manager or Linux Secret Service. Keys do not enter chat.
Is there guidance for a paid skill?
Yes, a monetization section: the free and paid split, landing page, payment and unobtrusive update notices.
Discussion
No questions or reviews yet.
Related
SkillsA skill for creating skills that proves them by behavior: before-and-after runs, regression evals and ecosystem audit
A skill that designs, tests, evaluates and packages other agent skills, starting from architecture
A skill that helps an agent author other skills to the agentskills.io spec: structure, metadata, progressive disclosure and validation
A doctor skill for agent skills that picks an architecture, audits against 21 red lines and returns a report with line references
A meta-skill that takes a vague idea to a finished agent skill: it designs, writes the files, checks them and refines by evidence
The `askit` command line toolkit for agent skills: lint, security scan, tests, forging new ones and export to any agent.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| oil-skill-creator (this card) | no | 102 | |
| skill-creator | no | 2,295 | turns an OpenAPI, GraphQL or MCP link into a ready skill; here review, trigger scoring and improvement of any skill |
| Functional Skill Creator | no | 256 | splits a long `SKILL.md` into functions with tests; here usefulness checks, boundaries and comparison with a plain agent |
README in short
The README presents the skill as a tool for those who treat a skill as a product: from judging usefulness to checking installation. The README says that a passing static check does not prove a skill is useful.