kata-from-github-issue
Use when a new coding exercise - a kata - shall be crafted from a GitHub issue.
Skill by wonderbird
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/wonderbird-kata-from-github-issue
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/kata-from-github-issue
What's inside
- Skill name
kata-from-github-issue- Description
- Use when a new coding exercise - a kata - shall be crafted from a GitHub issue.
- License
- GPL-3.0
- Instructions
- 100 lines in SKILL.md · view on GitHub
Before installing
- Instructions call the network
Files (1)
- SKILL.md
SKILL.md (first lines)
Please act as a software engineer skilled in quality assurance, requirements
engineering and technical writing.
Your goal is to create a feature specification from an existing issue in
GitHub.
The feature specification shall serve as exercise description for a coding
exercise. Target audience are developers new to the project. The goal of the
exercise is to learn agentic coding in a very large codebase.
## Workflow
1. If the user supplied the issue link when invoking this skill, use it.
Otherwise, ask the user for the issue link.
2. Get the issue description and all comments from GitHub, for example with
`gh issue view <number> -R <owner>/<repo> --comments`.
3. Find the pull request that fixed the issue, its merge commit and the
parent of the merge commit. Read the diff and the review comments.
If no merged fix exists, use the current default branch as starting point
and omit the reference solution.
4. Check every observable detail you intend to use in the acceptance
criteria against the code at the starting commit, e.g. exit codes, log
messages, endpoints, configuration keys and existing behavior. Do not
rely on memory: wrong details make the kata unsolvable.
5. Ask the user up to five questions, one by one. Ask only about points the
issue, the code and the defaults below do not settle, e.g. unclear scope,
non-goals or conflicting statements in the issue discussion.
6. Ask the user for the name of the kata owner for the license attribution.
Suggest the output of `git config user.name` as default. This question
does not count toward the five questions of step 5.
7. Write the specification to `README.md` in the repository root, using the
structure below. If `README.md` already exists, ask the user before
replacing it.
8. If `LICENSE` does not exist in the repository root, download the
CC BY-SA 4.0 legal code unchanged:
`curl -fsSL https://creativecommons.org/licenses/by-sa/4.0/legalcode.txt -o LICENSE`.
9. If you can build the project, build the reference solution and run the
acceptance criteria against it. Fix every criterion that fails.
If you cannot build it, tell the user which criteria are not validated.
10. Report the starting commit, the reference solution and the validation
status to the user.
## Defaults
Apply these defaults unless the user asks for something else.
- **Starting point:** the parent of the merge commit of the fix. The kata
then runs on the real codebase without the fix, and the upstream fix
serves as reference solution.
- **No hints:** describe required behavior only. Never name files,
functions, packages or code locations, and never reveal the approach of
the fix. Finding the relevant code is part of the exercise.
- **Functional scope only:** do not prescribe a development process or
agentic coding workflow.
- **Acceptance criteria follow project conventions:** observable details
such as exit codes and log message style mirror how the project already
handles comparable cases. Require automated tests for the new behavior.
## Specification Structure
Use these sections in this order:
1. Title: `# Kata: <capability>`, followed by a link to the issue.
2. `## Purpose of This Kata`: audience and learning goal.
3. `## Starting Point`: commit hash and subject, clone and branch commands,
required toolchain version.
4. `## Background`: domain concepts a newcomer needs, and the incident or
motivation from the issue.
5. `## Problem Statement`
6. `## Goal`
7. `## Non-Goals`: behavior that must stay unchanged, with the reason.
8. `## User Stories`
9. `## Functional Requirements`: numbered `### FR-<n>: <title>`.
10. `## Acceptance Criteria`: numbered `### AC-<n>: <title>`, written as
Given/When/Then with concrete inputs and observable outcomes.
11. `## Definition of Done`
12. `## Reference Solution`: link to the fixing pull request and its merge
commit, with the advice to compare only after finishing the kata.
Omit this section if no merged fix exists.
13. `## License`: see below.
## License Section
New katas are licensed under CC BY-SA 4.0. The `## License` section
contains:
- The statement that the kata is licensed under the
[Creative Commons Attribution-ShareAlike 4.0 International License](https://creativecommons.org/licenses/by-sa/4.0/)
(CC BY-SA 4.0), with a link to `LICENSE` for the full legal code.
- An attribution line for people who share or adapt the kata:
`"<kata title>" by <owner>, <repository URL>, licensed under CC BY-SA 4.0.`
Derive the repository URL from `git remote get-url origin`, converted to
an `https://` URL without the `.git` suffix. If there is no remote, ask
the user for the URL.
- One note per excerpt copied from the upstream project, e.g. code, rule
files or configuration snippets. The note names the upstream project, its
license and its copyright holder, because these excerpts stay under the
upstream license. Check the upstream license in the upstream repository.
Do not name file paths in the note, because the kata gives no hints.From wonderbird/ai-agent-workspace, GPL-3.0.
More from wonderbird/ai-agent-workspace
- 0.598score
record-findings
wonderbird
Use when you want to keep track of important follow-up tasks and findings that arise during the software development and review process.
SkillAFreshness: A★ 2Claude Code +9 - 0.589score
commit
wonderbird
Prepares and executes a git commit
SkillAFreshness: A★ 2Claude Code +9 - 0.563score
conductor
wonderbird
Run a working session as a conductor who delegates every task to subagents and never implements anything personally.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
memory-create
wonderbird
Creates a focused memory bank for the current project iteration using scope, git history, and tests.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
memory-to-docs
wonderbird
Use when learnings in the memory bank should be preserved beyond the current development focus by moving them to more permanent documentation.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
memory-update
wonderbird
Update memory bank files to keep them accurate for the active development.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
my-omc-parallel-work
wonderbird
Spawn parallel sub-agents in isolated git worktrees for independent Beads tasks.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
my-omc-show-next
wonderbird
Triage open issues down to the single next action.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
fix-problem
wonderbird
Remediation protocol for resolving obstacles one at a time via hypothesize, test, fix, and document.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
format-docs
wonderbird
Apply document style rules to markdown files such as meeting minutes, confluence pages, and project documentation.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
memory-bank-by-cline
wonderbird
Memory bank concept and structure for maintaining project knowledge across sessions.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
code-review
wonderbird
Check code quality according to best practices and standards.
SkillBFreshness: B★ 2Claude Code +9
Signals
- Stars
- 2
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 9, 2026 (20 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- GPL-3.0
- Repository
- github.com/wonderbird/ai-agent-workspace
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from wonderbird
- 0.598score
record-findings
wonderbird
Use when you want to keep track of important follow-up tasks and findings that arise during the software development and review process.
SkillAFreshness: A★ 2Claude Code +9 - 0.589score
commit
wonderbird
Prepares and executes a git commit
SkillAFreshness: A★ 2Claude Code +9 - 0.563score
conductor
wonderbird
Run a working session as a conductor who delegates every task to subagents and never implements anything personally.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
memory-create
wonderbird
Creates a focused memory bank for the current project iteration using scope, git history, and tests.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
format-docs
wonderbird
Apply document style rules to markdown files such as meeting minutes, confluence pages, and project documentation.
SkillBFreshness: B★ 2Claude Code +9 - 0.557score
fix-problem
wonderbird
Remediation protocol for resolving obstacles one at a time via hypothesize, test, fix, and document.
SkillBFreshness: B★ 2Claude Code +9
Cite this page
kata-from-github-issue skill. Agent Mart. https://agentmart.dev/skills/wonderbird-kata-from-github-issue (last updated 2026-10-24)