gh-discussion-replier

Gh Discussion Replier

Overview

The gh-discussion-replier skill enables posting professional English replies to GitHub Discussions. Unlike issues, discussions use the GraphQL API for both reading and posting. This skill handles the full workflow: fetching context, checking star status, composing a reply, and posting.

Workflow

1. Identify the Discussion: Get the discussion number (e.g., #66) and URL.
2. Fetch Discussion Content: Use GraphQL to get the title, body, author, and category.
3. Check Star Status: Run the bundled script to check if the author has starred the repo.
   • Command: bash .github/skills/gh-discussion-replier/scripts/check_discussion_star.sh <discussion_number>
   • Interpretation:
     • Exit code 0: User has starred. Use "Already Starred" templates.
     • Exit code 1: User has NOT starred. Include "Star Request" in the reply.
4. Select a Template: Load references/templates.md to choose a suitable response pattern based on the discussion category (Ideas, Q&A, General, Show and tell).
5. Draft the Reply: Compose a concise message based on star status and template.
6. Post the Comment: Use the GraphQL addDiscussionComment mutation (see below).

Tool Integration

Fetch Discussion Content

import subprocess, json

def get_discussion(owner, repo, number):
    query = '''
    query($owner:String!, $repo:String!, $number:Int!) {
      repository(owner:$owner, name:$repo) {
        discussion(number:$number) {
          id title body
          author { login }
          category { name }
          comments(first:10) { nodes { body author { login } } }
        }
      }
    }
    '''
    r = subprocess.run(['gh', 'api', 'graphql',
        '-f', f'query={query}',
        '-f', f'owner={owner}',
        '-f', f'repo={repo}',
        '-f', f'number={number}'],
        capture_output=True, text=True)
    return json.loads(r.stdout)['data']['repository']['discussion']

Check Star Status

bash .github/skills/gh-discussion-replier/scripts/check_discussion_star.sh <discussion_number>

Post Reply Comment

import subprocess, json

def post_discussion_reply(discussion_node_id, body):
    mutation = '''
    mutation($body:String!, $discussionId:ID!) {
      addDiscussionComment(input:{body:$body, discussionId:$discussionId}) {
        comment { id url }
      }
    }
    '''
    r = subprocess.run(['gh', 'api', 'graphql',
        '-f', f'query={mutation}',
        '-f', f'body={body}',
        '-f', f'discussionId={discussion_node_id}'],
        capture_output=True, text=True)
    return json.loads(r.stdout)

Full Workflow Example

import subprocess, json

OWNER = "Fu-Jie"
REPO = "openwebui-extensions"

# 1. Get discussion
disc = get_discussion(OWNER, REPO, 66)
disc_id = disc['id']
author = disc['author']['login']
category = disc['category']['name']
print(f"#{66} [{category}] by @{author}: {disc['title']}")

# 2. Check star
star_result = subprocess.run(
    ['bash', '.github/skills/gh-discussion-replier/scripts/check_discussion_star.sh', '66'],
    capture_output=True, text=True)
has_starred = star_result.returncode == 0

# 3. Compose reply (select template based on category and star status)
body = "Your reply here..."
if not has_starred:
    body += "\n\nIf you find this helpful, a star on [openwebui-extensions](https://github.com/Fu-Jie/openwebui-extensions) would be much appreciated! ⭐"

# 4. Post
result = post_discussion_reply(disc_id, body)
print(f"Posted: {result['data']['addDiscussionComment']['comment']['url']}")

Discussion Categories & Tone

Guidelines

• Language: ALWAYS use English for the comment body, even if the system prompt or user conversation is in another language.
• Tone: Warm, professional, and community-oriented. Discussions are conversations, not tickets.
• Precision: When announcing a fix or feature, mention the specific version and briefly describe what was added.
• Star Request: Always include a gentle star request if the user has NOT starred the repo. Skip it if they already have.
• Irreversibility: Discussion replies cannot be deleted via CLI. Always confirm the reply content with the user before posting.