| name | irasutoya-search |
| description | Search Irasutoya illustrations through this project's irasutoya CLI. Use when Claude needs to find Irasutoya/irasutoya.com illustrations, run irasutoya search, download or reuse the local CLI binary, summarize Irasutoya search results, or explicitly open/preview returned images. |
| argument-hint | <query> |
Irasutoya Search
Use the bundled script to perform real searches instead of reimplementing the scraper or guessing URLs.
Quick Start
Run the search script from this plugin skill directory:
python "${CLAUDE_SKILL_DIR}/scripts/irasutoya_search.py" $ARGUMENTS
If the user gives a query in natural language rather than slash-command arguments, pass that query to the same script:
python "${CLAUDE_SKILL_DIR}/scripts/irasutoya_search.py" cat
python "${CLAUDE_SKILL_DIR}/scripts/irasutoya_search.py" "office worker"
Open images only when the user explicitly asks to open or preview them:
python "${CLAUDE_SKILL_DIR}/scripts/irasutoya_search.py" --open-images dog
The summary shows up to five image URLs per result by default. Use --all-images only when the user asks for every image URL.
Behavior
- Resolve
irasutoya on PATH first.
- If PATH resolution fails, reuse a cached downloaded binary from
scripts/bin/<tag>/<os>_<arch>/.
- If no cache exists, query the latest GitHub release for
Mineru98/irasutoya-cli, download the matching OS/arch archive and checksums.txt, verify the archive SHA-256, extract only inside scripts/bin/<tag>/<os>_<arch>/, then run it.
- Keep default output text-only. Pass
--open-images to the CLI only for explicit image-open requests.
- If no published release exists, report the blocker. Do not claim live release fallback succeeded.
Validation
Use deterministic checks before trusting edits to the script:
python "${CLAUDE_SKILL_DIR}/scripts/test_irasutoya_search.py"
