About
This Claude Skill handles GitHub interactions specifically for the `Necmttn/ax` repository, allowing users to star it, file issues, or open pull requests via the `gh` CLI. It triggers only on explicit user requests for ax-related actions and always confirms before mutating account state. If the CLI is unavailable, it gracefully falls back to providing direct GitHub URLs.
Quick Install
Claude Code
Recommendednpx skills add Necmttn/ax -a claude-code/plugin add https://github.com/Necmttn/axgit clone https://github.com/Necmttn/ax.git ~/.claude/skills/ax-repoCopy and paste this command in Claude Code to install this skill
Documentation
ax:repo
Let an AI coding agent interact with the ax GitHub repo (Necmttn/ax) for the
user - star, file an issue, or fork + open a PR - without the user
ever typing a gh command. Everything routes through the already-installed
gh CLI; there is no axctl surface for this.
Repo: Necmttn/ax · https://github.com/Necmttn/ax
When to fire
- "star ax" / "star the repo" / "I want to support ax" / "give ax a star"
- "report this as an ax bug" / "file an ax issue" / "open an issue on ax"
- "this looks like an ax bug" (after an
ax/axctlerror) - "I want to contribute to ax" / "fix this in ax" / "open a PR against ax"
Do NOT fire for GitHub work on other repos or for general gh usage.
Issues and PRs are user-initiated only. Star is the exception - you may
offer it proactively (see Proactive star nudge), but offering ≠ doing: the
actual star always needs an explicit yes.
Non-negotiable rules
- Preflight
ghfirst (read-only, no confirm). Detect three states:ghmissing → fall back to a plain URL (see Fallback). Don't error.ghpresent but unauthenticated → fall back to a plain URL, and tell the user they cangh auth loginto do it inline next time.ghpresent + authed → proceed to the action (still confirm mutations).
- Confirm before any account-mutating call - star, issue create, PR
create all change the user's GitHub account/repo. Show the exact command,
get an explicit yes. Read-only checks (
gh auth status, "is it starred") need NO confirm. - Never break the caller's exit code, never run silently in CI. If
$CIis set or stderr/stdin isn't a TTY, do NOT run mutatingghcalls - print the URL instead. A failedghcall must not abort the user's task.
Preflight (run this before every action)
if ! command -v gh >/dev/null 2>&1; then
echo "GH_STATE=missing"
elif ! gh auth status >/dev/null 2>&1; then
echo "GH_STATE=unauthed" # gh auth status exits 4 when not logged in
else
echo "GH_STATE=ok"
fi
missing/unauthed→ use Fallback for the chosen action.ok→ use theghcommand for the chosen action (after confirm).
Actions
1. Star
Mutates the user's account → confirm first.
# Optional read-only check (no confirm): is it already starred?
# exits 0 (starred) / non-zero / 404 (not starred)
gh api /user/starred/Necmttn/ax >/dev/null 2>&1 && echo "already starred"
# The star (after explicit yes):
gh api -X PUT /user/starred/Necmttn/ax # silent 204 on success
If already starred, say so and skip - don't re-PUT or re-prompt.
Unstar (only if asked): gh api -X DELETE /user/starred/Necmttn/ax.
You can also just run ax star, which stars via gh (or prints the URL) and
silences the CLI's periodic star reminder.
Fallback (gh missing/unauthed): print
https://github.com/Necmttn/ax and tell the user to click Star.
Proactive star prompting is NOT this skill's job - the CLI handles it deterministically (a once-a-day stderr footer shown only on an interactive terminal until the user runs
ax star/ax star --done). This skill only acts on an explicit user request.
2. File an issue / bug report
Mutates (creates an issue) → confirm first, and show the title/body you'll submit so the user can edit before you send it.
Interactive (opens a prefilled browser form - good default when the user wants to review/edit in GitHub's UI):
gh issue create --repo Necmttn/ax --web
Non-interactive (prefilled title + body, e.g. an error report you assembled):
gh issue create --repo Necmttn/ax \
--title "<concise summary>" \
--body "<body>" \
--label feedback # only if the user confirms; omit if unsure label exists
Error-report pattern. When firing off the back of an unhandled ax/axctl
error, prefill from the failure - never make the user paste a stack trace:
gh issue create --repo Necmttn/ax \
--title "ingest: <one-line error>" \
--body "$(cat <<'EOF'
**Command:** `ax <subcommand> <args>`
**ax version:** <output of `ax --version`>
**OS:** <uname -srm>
**What happened**
<one or two sentences>
**Error**
<the actual error output - trimmed, no secrets>
EOF
)"
Scrub paths/tokens that might leak private data before submitting. Confirm the assembled body with the user first.
Fallback (gh missing/unauthed): print the web new-issue URL. You can
prefill it via query string:
https://github.com/Necmttn/ax/issues/new?title=<urlencoded>&body=<urlencoded>
(plain https://github.com/Necmttn/ax/issues/new also works). Tell the user to
review and submit in the browser.
3. Fork + open a PR (contribute)
For a code change. Fork+clone is account-mutating → confirm before the fork and before the PR; branching/committing locally needs no confirm.
# 1. Fork and clone in one step (creates a fork on the user's account):
gh repo fork Necmttn/ax --clone # confirm: this creates a fork
# 2. From inside the clone, branch + make the change + commit:
git checkout -b <topic-branch>
# ...edits...
git commit -am "<conventional message>"
git push -u origin <topic-branch>
# 3. Open the PR against upstream (confirm before sending):
gh pr create --repo Necmttn/ax \
--title "<title>" --body "<what + why>"
# or interactively review in browser:
gh pr create --repo Necmttn/ax --web
If the user is already inside a clone of Necmttn/ax, skip the fork step;
gh pr create will offer to push to a fork automatically.
Fallback (gh missing/unauthed): print https://github.com/Necmttn/ax/fork
and tell the user to fork in the browser, then clone their fork manually.
House rules
- Show the exact
ghcommand before running any mutating one; get a yes. - One action per request - don't star and file an issue unless asked for both.
- Don't invent labels/milestones; omit
--labelif you're unsure it exists. - Keep issue bodies short, factual, secret-free; never paste raw transcripts.
- On any
ghfailure, surface the error and offer the URL fallback - never let it abort the user's in-progress task.
GitHub Repository
Frequently asked questions
What is the ax-repo skill?
ax-repo is a Claude Skill by Necmttn. Skills package instructions and resources that Claude loads on demand, so Claude can perform ax-repo-related tasks without extra prompting.
How do I install ax-repo?
Use the install commands on this page: add ax-repo to Claude Code as a plugin, or clone its repository into your skills directory, then restart Claude so it picks up the skill.
What category does ax-repo belong to?
ax-repo is in the Meta category, tagged ai.
Is ax-repo free to use?
Yes. ax-repo is listed on AIMCP and free to install.
Related Skills
This skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
This skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
This skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
