Menu
Reference for gh CLI commands—auth, repo context, GraphQL conventions. Workflow skills delegate here for shared mechanics.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
| name | use-github |
| description | Reference for gh CLI commands—auth, repo context, GraphQL conventions. Workflow skills delegate here for shared mechanics. |
Reference for shared gh CLI mechanics—auth, repo context, and GraphQL conventions. Workflow skills own their specific commands and delegate here for these common patterns.
gh needs required_permissions: ["full_network"].
gh auth status
If failed: Tell user gh auth login.
gh repo view --json owner,name
Extract owner.login and name — needed for REST and GraphQL API calls.
gh api graphql \
-f query="$(cat path/to/query.graphql)" \
-F owner=<owner> \
-F name=<repo> \
-F number=<number>
Pass each variable as its own -F flag (not a variables object). Use -f (lowercase) for string values and -F (uppercase) for typed values (integers, booleans).
Each skill owns its own .graphql query files. Follow these conventions:
pageInfo { hasNextPage endCursor } on paginated connections. If hasNextPage is true, tell the user that more results exist and offer three options: fetch the next page, fetch all remaining pages, or continue with the current page only.databaseId on any node you will reference via the REST API (REST endpoints use integer IDs, not GraphQL node IDs).createdAt on comments when ordering matters.When pageInfo.hasNextPage is true, fetch the next page by passing the endCursor value as the after argument on the connection:
gh api graphql \
-f query="$(cat path/to/query.graphql)" \
-F owner=<owner> \
-F name=<repo> \
-F number=<number> \
-f after=<end_cursor>
Query templates that support pagination should accept an optional $after: String variable and pass it to the connection (e.g. reviewThreads(first: 50, after: $after)). The $after is nullable so the first call works without it.
Merge results across pages before processing.
gh auth login.Based on SOC occupation classification
Generate platform-tailored launch posts for an open-source project. Analyzes the current repo (git log, README, key source files), web-searches for relevant communities (subreddits, HN, Lobsters, Discord servers, etc.), drafts a title + body per platform matching that community's tone, runs each body through the humanizer skill, and writes a single LAUNCH_POSTS.md at the repo root. Use whenever the user wants to announce a new repo, share their project, draft a Show HN, write subreddit launch posts, or otherwise promote OSS work — even when they only mention one platform, since this produces a multi-platform set in one shot. Triggers on phrases like "launch posts", "Show HN draft", "announce my project", "subreddit posts", "promote my OSS", "how do I share this on Reddit/HN", "where should I post this", or "make me launch posts".
Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.
Save conversation knowledge, code explanations, research, and decisions to Obsidian as beautifully structured markdown notes. Use this skill whenever the user mentions Obsidian, wants to save/store/export notes to their vault, says 'save to obsidian', 'obsidian init', 'read from obsidian', or wants to persist conversation insights as permanent notes. Also trigger when the user wants to organize knowledge, create reference docs, or archive session work into their personal knowledge base — even if they don't say 'Obsidian' explicitly but have previously initialized a vault.
