Guide · Agent skill
How to use the Release Publisher agent skill (release-publish)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The release-publish skill is an example agent skill from the ai-engineering-from-scratch learning project, under the MIT licence. It gives an AI agent a short instruction set for publishing a software release that someone has already approved. This guide explains what the skill tells the agent to do, how to install it, how to trigger it with a natural request, and where it falls short for real releases.
Release Publisherrohitg00/ai-engineering-from-scratch · MIT Download the skill Quick start
- 1Download the ZIP. Get the release-publish ZIP from the skill page on this site.
- 2Unzip the folder. Extract it so that the folder contains SKILL.md and routing.json.
- 3Place it where the agent looks. Put the folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder, or upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
- 4Make a request that matches the description. Ask the agent to publish an approved Orchard release, and include the release details and any evidence it should use.
- 5Check the output. Confirm the result lists source file paths and that the agent stopped where evidence was missing. Verify every step before anything goes live.
What is an agent skill?
An agent skill is a folder containing a SKILL.md file. The file has a short header with a name and a description, followed by instructions that an AI agent reads when a task matches the skill. The agent uses the description to decide when the skill is relevant.
This skill's SKILL.md has the name release-publish and the description Publish an approved Orchard release. Its body is only two sentences of guidance.
What does the release-publish skill make the agent do?
The skill sets two rules. First, the agent should keep source file paths in its result, so you can trace where each piece of the output came from. Second, the agent should stop when evidence needed for the release is missing, rather than guess.
The source does not define what Orchard is, and it does not list concrete publishing steps, commands, or checks. The skill describes a workflow pattern, not a working release pipeline.
- Keep source file paths in the result.
- Stop when evidence for the release is missing instead of guessing.
- Publish only a release that has already been approved.
How do you install the release-publish skill?
Download the ZIP from the skill page on this site and unzip it. You can install the skill in one of three ways.
For a personal install, unzip the folder into ~/.claude/skills/. To scope it to one project, unzip it into that project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
- Personal: place the unzipped folder in ~/.claude/skills/
- Project: place the unzipped folder in your project's .claude/skills/
- Claude apps: upload the ZIP under Settings → Capabilities → Skills
How do you trigger the skill with a natural request?
The agent decides whether to use a skill based on its description, so a request that matches the description's wording is the most likely to trigger it. The description mentions publishing an approved Orchard release, so a request like the one below is a reasonable starting point.
Be explicit that the release is already approved. The skill's own text assumes approval has happened, and it tells the agent to stop if evidence is missing, so give the agent the material it needs in the same request.
- Publish the approved Orchard release for version 2 using the notes in CHANGELOG.md.
- Publish the approved Orchard release and list the source paths you relied on.
- Publish the approved release, and stop if any evidence is missing.
What does the skill need and what are the safety notes?
The source lists no requirements. It does not run scripts and does not use the network, according to the listing. The package contains two files: routing.json and SKILL.md.
Because the SKILL.md gives no concrete steps, commands, or checks, an agent using it for a live release would have to work out the process on its own. For production releases, that is the main risk. Treat the agent's output as a draft and verify every step yourself before anything is published.
- Requires: nothing listed
- Runs scripts: no
- Uses network: no
- Files: routing.json, SKILL.md
Who should use this skill, and what should you use instead?
This skill is most useful as a minimal example. Developers can read it and adapt the structure for their own pipeline, and teams studying skill routing can see how a skill file is laid out.
For actual production releases, use a release process that your team has written and tested, with explicit commands and checks. The skill does not define any publishing system, so you will need to supply that yourself.
- Good fit: adapting an example for your own pipeline, or studying skill routing files.
- Poor fit: publishing production releases without a documented process.
What people use it for
- Reading the SKILL.md as a template for writing your own release-publish style skill.
- Studying how a routing.json file sits alongside SKILL.md in a skill folder.
- Demonstrating to a team how an agent can be told to stop instead of guessing when evidence is missing.
Frequently asked questions
Is the release-publish skill free to use?
The source is licensed under the MIT licence, which permits use, modification and redistribution. The listing does not mention any price.
What do I need to run the release-publish skill?
The listing states no requirements. It runs no scripts and uses no network, so it adds instructions to the agent rather than tools to install.
Can this skill publish a production release for me?
Not reliably. The SKILL.md has no concrete publishing steps, commands, or checks, so the agent would have to invent the process. Use a tested release process for production work.
What is Orchard?
The files do not define Orchard. The skill refers to a release for Orchard, but the system itself is not described in the SKILL.md or the listing.
Where do I put the skill so Claude can find it?
Unzip the folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
How is this different from other release tools?
This is a short instruction file, not a tool. Release automation tools run commands and checks themselves, whereas this skill only tells an agent how to behave: keep source paths and stop when evidence is missing.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use release-publish if you want a minimal, readable example of an agent skill to learn from or adapt for your own pipeline. Do not rely on it to publish production releases, because it contains no concrete steps or checks and the agent would have to fill those gaps itself; choose a documented, tested release process instead.
Download the skill