find-issueslisted

Analyze the repository at `$REPO_URL` and find open issues that match your profile. ## Your Profile (from config.json) Profile information is read from `~/.claude/config.json`: - **Website**: From `profiles.website` - **Blog**: From `profiles.blog` - **GitHub**: From `profiles.github` ## Step 1: Discover Interests (Single Parallel Call) Launch these two MCP calls **in one parallel batch** — nothing else: ``` mcp__github__get_me() mcp__github__search_repositories(query="user:{username}", sort="updated") ``` Extract from results: - Bio, languages, topics from repos - Primary tech stack (infer from top 10 most recent repos) **Do NOT** fetch website/blog — it's slow and rarely adds signal beyond GitHub data. ## Step 2: Fetch Issues (Use GraphQL API, NOT Search API) **CRITICAL**: Use `mcp__github__list_issues` (GraphQL), NOT `mcp__github__search_issues` (REST). The REST search API triggers **secondary rate limits (403)** when making multiple parallel calls. GraphQL `list_issues` is more reliable and supports server-side filtering. ``` mcp__github__list_issues( owner="{owner}", repo="{repo}", state="OPEN", since="6 months ago ISO 8601", direction="DESC", orderBy="UPDATED_AT", perPage=50 ) ``` **If the result is too large** (>50KB), delegate filtering to a **sonnet subagent** with this prompt pattern: > "Read the file at {path} in chunks. Find issues that are: unassigned, no linked PRs, > have labels like bug/feature-request/documentation/open-to-a-pull-requ