Add New Entry
Formaterar råa URL:er till legend-stylade wiki-poster för azure-openai-llm-wiki-repot
Fungerar med konfiguration
Vad den gör
En repo-maintenance skill som omvandlar en raw temp.md checklist of URLs into properly formatted markdown entries for the azure-openai-llm-wiki section files, applying per-file legend rules (date format, star badges, emoji symbols, section names, 15-word descriptions). Utlöses when adding new resources to that knowledge base; explicitly not for editing existing entries or restructuring sections. Relies on Python helper scripts under the repo's code/ directory to fetch GitHub descriptions, creation dates, and star counts.
Testrapport
Hittade SKILL.md på .agent/skills/add-new-entry/SKILL.md; frontmatter har name+description. Punktkontrollerade de 3 refererade skripten: 404 på den skill-relative path men 200 på repo-root code/ (fetch_github_description.py, get_github_dates.py, add_github_stars.py) — the skill is coupled to the repo root, not self-contained. Inga security smells. För OUTPUT matade jag in en 2-line temp.md (a GitHub repo + an arXiv paper) och producerade en baseline_entries.md (my natural formatting) vs skill_entries.md (following the legend rules). Konkreta measured differences: skill used dash+parens date `(Feb 2023)` and no emoji for the azure.md entry vs colon+brackets `[Jan 2025]` with the 📑 symbol appended inside `[DeepSeek-R1📑]` for the research entry; it put a real github-stars badge only on the GitHub line while baseline wrongly slapped ⭐ on the arXiv line too; descriptions dropped to 9 and 6 words vs baseline's verbose, name-repeating text. Baslinjen hade no dates and identical formatting for both. The arXiv date I derived from the ID rule (2501→Jan 2025); the GitHub creation date would need the script to confirm.
Testad: 2026-07-21 · Claude Code 2.x (agent harness)
Installation
git clone --depth 1 https://github.com/kimtth/azure-openai-llm-wiki.git /tmp/add-new-entry-src mkdir -p ~/.claude/skills cp -R /tmp/add-new-entry-src/.agent/skills/add-new-entry ~/.claude/skills/add-new-entry # REPO-COUPLED: the skill body runs `python code/<script>.py` (fetch_github_description.py, # get_github_dates.py, add_github_stars.py) and edits section/*.md — these live in the cloned # repo ROOT, not in the skill folder, so a standalone skill-dir install is incomplete. # To actually run the workflow, work from inside /tmp/add-new-entry-src (or your own clone) # where code/ and section/ exist, and create a local temp.md input file first. # Deps: Python 3 + network access to GitHub and Semantic Scholar APIs.
Kommandon och exempelprompter
/add-new-entryFormaterar råa URL:er till legend-stylade wiki-poster för azure-openai-llm-wiki-repot
Skills triggas av vanliga förfrågningar — inga kommandon att memorera. Efter installationen aktiverar prompter som dessa skillen (på engelska):
Add this new resource from temp.md to a sectionFile this entry into the correct wiki pageAdd a new tool entry to the knowledge base