How to use the Create GitHub Issues from Plans agent skill

What an agent skill is and what this one makes the agent do

An agent skill is a folder containing a SKILL.md file with instructions that an AI agent loads when a task matches the skill. This skill is only instructions; it does not run code itself.

When you point the agent at an implementation plan, the skill asks it to analyze the plan to identify phases, check existing issues with search_issues, then create a new issue per phase with create_issue or update an existing one with update_issue. It asks the agent to use the repository's feature_request.yml or chore_request.yml templates, falling back to the default if those are not present.

• One issue per implementation phase
• Issue title is the phase name from the plan
• Issue description contains the phase details, requirements and context
• Only changes required by the plan are included
• Labels chosen to suit the issue type (feature or chore)

Install the skill from the ZIP download

Download the ZIP for this skill from this page and unzip it. You can install it for yourself across all projects or for a single project.

For a personal install, place the unzipped skill folder in ~/.claude/skills/. For a project-only install, place it in the project's .claude/skills/ folder.

Alternatively, in the Claude apps you can upload the skill under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/
• Project only: .claude/skills/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload

Check your requirements before you start

The skill assumes a few things are in place. Without them the agent cannot complete the task.

The skill assumes GitHub issue tools (search_issues, create_issue, update_issue) are available to the agent, so your project must be hosted on GitHub. You also need a GitHub account with access to the target repository and an implementation plan file with identifiable phases.

• A GitHub account with access to the target repository
• GitHub issue tools available to the agent: search_issues, create_issue, update_issue
• An implementation plan file with identifiable phases

Trigger the skill with a natural request

The skill is written to act on the plan file you reference. Name the plan file and say what you want done with its phases. The skill's instructions refer to the plan at the file you supply, so make sure the agent has that file open or can read it.

Good requests name the plan and the outcome, for example asking the agent to create GitHub issues for each phase of a named plan file, or to update issues that already exist for those phases.

• "Create GitHub issues for each phase in docs/plans/payments-plan.md."
• "Check whether any of the phases in this plan already have issues, and update those instead of duplicating them."
• "Turn the phases of this implementation plan into feature issues in our repo."

Review what the agent creates

Because the skill creates and modifies issues directly, check the results in GitHub after the run. Confirm that each phase became one issue, that no duplicates were created, and that each issue contains only what the plan requires.

The skill is not meant to write to a shared tracker without review. If your team needs sign-off first, ask the agent to list proposed titles and bodies before creating anything, or review the output yourself.

• Check for one issue per phase
• Check for duplicates against existing issues
• Check labels and that the feature or chore template was used

Safety notes: scripts and network access

According to the skill's listing, it runs no scripts. It does use the network, because its work happens through GitHub issue tools that read and write to your repository.

Because it modifies issues in a live repository, use a GitHub account and permissions you are comfortable letting an agent use, and review changes on shared trackers.

• Runs scripts: no
• Uses network: yes, through GitHub issue tools
• Writes to issues directly, so review on shared repositories