Agent Skills are folders with instructions and resources that Claude and Codex load when relevant. Turn a process you repeat into a skill, test it, and share it with your team.
Independent review. Unpaid; no partnership with the tools on this page. Based on public documentation and research. Trademarks identify their owners’ products. Disclosure.
Editorial shortlist
Compare your options.
For maintaining reusable agent instructions and tests.
Verify the target harness; skill files are not automatically portable. Choose one for implementing and reviewing code changes; switching may need setup changes.
Editorial judgment, not a personal-use endorsement or performance score. These are alternatives for the named role, not three required stack components. Vendors did not pay for, request or approve these positions.
Inspectable file changes come before editor convenience. A lower-ranked option may fit your constraints better; compare the conditions stated on each card before choosing.
Setup and portability: Use scoped repositories, tests and human review. Switching agents does not migrate settings, credentials or model access.
Reviewed 2026-10-21. This page applies a shared comparison for “implementing and reviewing code changes” to the stated task and constraints. It is documentation-based editorial guidance, not three new hands-on tests. We weigh value for the job, distinct capabilities, replacement limits, technical effort and interoperability. Check current plan terms before adopting or switching.
Review skill outputs and update the skill, not just the output
Human stepConfirmed
Confirmed = vendor-documented, not Ion-tested. Human and AI-platform steps are proposed roles.
Human approval
You review outputs and own the skill contents.
Watch out for
A skill that encodes a weak process scales the weakness. Test it on three real tasks before you share it, and keep the master copy in GitHub, not in a chat.
Starter prompt
Here is a process I repeat: [describe it]. Draft a SKILL.md folder for it: a short description of when to use it, the steps, two worked examples and any script it needs. Keep it tool-neutral where you can. I will review the skill, test it on a real task and commit it to our skills repository through a pull request.
Written for this workflow. Paste it into your AI platform and adjust it to your situation.
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
Gemini
Confirmed
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test.
What the platform statuses mean
Confirmed: documented access meets the recorded requirements. Inferred: a reviewed note describes a plausible, untested pairing. To verify: a connection or step is unresolved. MCP and API pairings need route-specific evidence; a plugin, CLI or API alone does not prove compatible access, credentials or permissions. These are documentation assessments, not IonBin runtime tests.
What we checked (2026-10-18)
The format. A skill is a folder with a SKILL.md file that has at least a name and a description, plus optional scripts and reference files. Agent Skills is published as an open format at agentskills.io.
Claude. Custom skills work in Claude Code (~/.claude/skills/ for personal skills, .claude/skills/ for a project), in claude.ai settings and through the Claude API. Claude Code’s current guide documents one-way downloads of enabled claude.ai account skills for supported signed-in sessions. Local-only skills do not automatically reach Cowork or cloud routines, and API uploads remain separate. Check the surface and sign-in method rather than assuming all copies sync. It also says to use skills only from sources you trust, because a skill can contain code.
Codex.Codex reads skills from .agents/skills folders, from your working directory up to the repository root, as well as user and admin locations. It loads a skill’s full instructions only when it picks the skill. In Codex CLI and the IDE extension, you can call one with $ or /skills; ChatGPT uses @. Plugins are the documented distribution route for reusable skills across ChatGPT and Codex.
Portability. Both use SKILL.md, but the folder locations differ. Keep one authoritative source and deliberately install or distribute it to each supported surface. Test scripts, dependencies, permissions and vendor-specific fields in each client.
Set it up
Choose a task you’ve explained more than three times.
Write the steps, one good example and the common mistakes.
Test it on a real task in each agent.
Store the library in a GitHub repository, so changes get reviewed.
Verification limits and implementation checks
The page is documentation-reviewed; the complete workflow has not been run by IonBin.
Verify the skill on each target client; sharing SKILL.md does not guarantee identical tools or behavior.
Check a skill before you install it
Sentry’s Skill Scanner reviews a third-party skill for injection, risky scripts, broad permissions and outside links before you add it. It runs its own script and asks for Bash, so use it in a disposable environment, and still read any scripts yourself. IonBin reviewed it from source only.