Skip to content

uithub.md

Latest commit

 

History

History
21 lines (19 loc) · 1.76 KB

uithub.md

File metadata and controls

21 lines (19 loc) · 1.76 KB
Raw

Purpose: fetch UiThub repo context for a GitHub project without cloning.

Inputs

  • $REPO_URL (required): GitHub URL or owner/repo slug.
  • $CONTEXT (optional): focus areas/questions to guide which files to fetch.

Flow

  1. Validate $REPO_URL. If missing or invalid, ask the user for a correct value.
  2. Normalize to <owner>/<repo> and extract any path hints from $REPO_URL.
  3. Determine the repository’s default branch (e.g., curl -H "Accept: application/vnd.github.v3+json" https://api.github.com/repos/<owner>/<repo> | jq -r '.default_branch'). If discovery fails, fall back to main but mention the assumption.
  4. Fetch structure only:
    curl "https://uithub.com/<owner>/<repo>?accept=text/plain&omitFiles=true" > tree.txt
  5. Pull key docs (adjust lines) using the discovered branch:
    curl "https://uithub.com/<owner>/<repo>/tree/$BRANCH/README.md?accept=text/plain&plain=1&lines=200" > readme.txt
    • Fetch other docs (CONTRIBUTING.md, docs/overview.md, etc.) when $CONTEXT suggests.
    • After each curl, check the output (status code, byte size). Retry with plain=1, ?path=..., or https://raw.githubusercontent.com/<owner>/<repo>/$BRANCH/... if the response is empty or a 404.
  6. Capture code slices guided by $CONTEXT:
    • Start with curl "https://uithub.com/<owner>/<repo>/tree/$BRANCH/<path>?accept=text/plain&plain=1&lines=200" for targeted files.
    • For larger sections, use curl "https://uithub.com/<owner>/<repo>?accept=text/plain&path=<path>", then trim locally.
    • If UiThub still returns nothing, fall back to https://raw.githubusercontent.com/<owner>/<repo>/$BRANCH/<path>.
  7. Summarize what each artifact reveals and note gaps needing further UiThub pulls or a full clone, explicitly tying the findings back to $CONTEXT.