arr-search
Search radarr/sonarr/sportarr by title via /api/v3/{movie|series}/lookup using the X-Api-Key header. Read-only. Returns up to 5 matches with poster URLs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Search radarr/sonarr/sportarr by title via /api/v3/{movie|series}/lookup using the X-Api-Key header. Read-only. Returns up to 5 matches with poster URLs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
Poll trusted Multiple Sclerosis RSS feeds, score credibility, save raw items, filter low-quality material, and promote high-value items into citation-grounded KB processing.
Save a URL, file path, pasted text, DOI, or PMID into the knowledgebase at $KB_ROOT/raw/manual/YYYY/MM/DD/. Enriches via the relevant MCP (pubmed/crossref) when input is a DOI or PMID. NON-DESTRUCTIVE — write-only into raw/, never into pages/.
Append a single timestamped bullet to today's $KB_ROOT/journals/YYYY/MM/YYYY_MM_DD.md describing what the agent just did. Logseq-compatible and date-partitioned. Call this at the end of every meaningful action (page write, ingest, query answered).
Self-heal the MS knowledgebase layout after RSS watch or weekly digest. Migrates legacy flat files into the V2 date/type structure, refreshes Start Here and Index pages, checks links/frontmatter, and journals the maintenance summary.
Answer a research question by searching the kb first, then PubMed → CrossRef → SearXNG, writing a citation-grounded content page under $KB_ROOT/content/<type>/YYYY/MM/. Every claim cites a DOI/PMID/URL or is dropped. This is the heart of the agent.
Safely prune generated RSS watcher raw cache after weekly digest retention, without touching manual raw ingests or the RSS seen ledger.
| name | arr-search |
| description | Search radarr/sonarr/sportarr by title via /api/v3/{movie|series}/lookup using the X-Api-Key header. Read-only. Returns up to 5 matches with poster URLs. |
| version | 0.1.0 |
| author | dobry-ops |
| license | MIT |
| metadata | {"hermes":{"tags":["media","search","radarr","sonarr","sportarr"],"category":"media-ordering","related_skills":["arr-library","qbit-list","nzbget-list"]}} |
Read-only title lookup against any *arr app in the homelab.
User asks to search/find/look-up a title on radarr (movies), sonarr (series), or sportarr (sports series). Examples:
$RADARR_URL / $RADARR_API_KEY$SONARR_URL / $SONARR_API_KEY$SPORTARR_URL / $SPORTARR_API_KEYMovies (radarr):
curl -fsS -H "X-Api-Key: $RADARR_API_KEY" \
--get --data-urlencode "term=<query>" \
"$RADARR_URL/api/v3/movie/lookup" \
| jq '.[0:5] | map({title, year, tmdbId, imdbId, remotePoster})'
Series (sonarr / sportarr — both use the same series shape):
curl -fsS -H "X-Api-Key: $SONARR_API_KEY" \
--get --data-urlencode "term=<query>" \
"$SONARR_URL/api/v3/series/lookup" \
| jq '.[0:5] | map({title, year, tvdbId, imdbId, remotePoster})'
Single markdown table. Cap at 10 rows unless the user asks for more. Posters, TMDb, and IMDb columns are LINKS (markdown link syntax), not inline images and not raw URLs.
| Pick | Title | Year | Runtime | TMDb | IMDb | In Radarr | Poster |
|---:|---|---:|---:|---:|---|---:|---|
| 1 | **Mulan** | 1998 | 88m | [10674](https://www.themoviedb.org/movie/10674) | [tt0120762](https://www.imdb.com/title/tt0120762) | no | [poster](https://image.tmdb.org/t/p/original/...jpg) |
Link templates:
https://www.themoviedb.org/movie/<tmdbId>https://www.themoviedb.org/tv/<tvdbId-or-tmdbId>imdbId is present): https://www.imdb.com/title/<imdbId>remotePoster URL from the API response, exactly as-isIf a field is missing in the API response (no imdbId, no remotePoster),
print n/a for that cell — don't emit a broken link.
After the table, one short paragraph with your "Saxton guess" (the 1–2 most
likely picks) and the next-step prompt: Say "inspect N" to see release candidates.
$RADARR_API_KEY). Never paste values into chat.