メニュー
Paperless-ngx document management — search, tag, organize, and retrieve documents
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/dapperdivers/roundtable-arsenal --skill paperless-opsコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Synthesize domain knight summaries into a comprehensive Daily Briefing for Derek's Obsidian vault. Your output IS the briefing file — write the full document, not a summary of it.
Procedure for the morning-briefing verify_claims step: re-run the check behind every mechanically-verifiable claim in the domain summaries, build the authoritative knight-liveness map from vault file mtimes, update the open-items ledgers, and emit one JSON verification report that the synthesizer treats as the only source of numbers.
Protocol for the per-domain open-items ledger at /vault/Briefings/Ledger/<domain>.json. Replaces prose memory as the carry-forward mechanism for tracked items: Day-N counters are computed from first_seen, items are re-verified before being carried, and resolutions are recorded machine-readably so they can never be silently lost.
SOC 職業分類に基づく
SKILL.md を表示中
| name | paperless-ops |
| description | Paperless-ngx document management — search, tag, organize, and retrieve documents |
Manage documents in Paperless-ngx via REST API.
http://paperless.selfhosted.svc.cluster.local:8000Authorization: Token $PAPERLESS_TOKENPAPERLESS="http://paperless.selfhosted.svc.cluster.local:8000/api"
AUTH="Authorization: Token $PAPERLESS_TOKEN"
# Full-text search
curl -s -H "$AUTH" "$PAPERLESS/documents/?query=tax+2025" | jq '.results[] | {id, title, created}'
# Filter by tag
curl -s -H "$AUTH" "$PAPERLESS/documents/?tags__id__in=5" | jq '.results[] | {id, title}'
# Filter by correspondent
curl -s -H "$AUTH" "$PAPERLESS/documents/?correspondent__id=3" | jq '.results[] | {id, title}'
# Filter by document type
curl -s -H "$AUTH" "$PAPERLESS/documents/?document_type__id=2" | jq '.results[] | {id, title}'
# Filter by date range
curl -s -H "$AUTH" "$PAPERLESS/documents/?created__date__gte=2025-01-01&created__date__lte=2025-12-31"
# Combine filters
curl -s -H "$AUTH" "$PAPERLESS/documents/?query=invoice&tags__id__in=5&ordering=-created"
curl -s -H "$AUTH" "$PAPERLESS/tags/" | jq '.results[] | {id, name, document_count}'
curl -s -H "$AUTH" "$PAPERLESS/correspondents/" | jq '.results[] | {id, name, document_count}'
curl -s -H "$AUTH" "$PAPERLESS/document_types/" | jq '.results[] | {id, name, document_count}'
curl -s -H "$AUTH" "$PAPERLESS/documents/42/" | jq '{id, title, content, tags, correspondent, document_type, created}'
# Original file
curl -s -H "$AUTH" "$PAPERLESS/documents/42/download/" -o document.pdf
# Thumbnail
curl -s -H "$AUTH" "$PAPERLESS/documents/42/thumb/" -o thumb.png
# Add tags (PATCH preserves other fields)
curl -s -X PATCH -H "$AUTH" -H "Content-Type: application/json" \
"$PAPERLESS/documents/42/" \
-d '{"tags": [5, 8, 12]}'
# Update title
curl -s -X PATCH -H "$AUTH" -H "Content-Type: application/json" \
"$PAPERLESS/documents/42/" \
-d '{"title": "2025 W2 - Employer Name"}'
# Set correspondent
curl -s -X PATCH -H "$AUTH" -H "Content-Type: application/json" \
"$PAPERLESS/documents/42/" \
-d '{"correspondent": 3}'
curl -s -X POST -H "$AUTH" -H "Content-Type: application/json" \
"$PAPERLESS/tags/" \
-d '{"name": "tax-2025", "color": "#ff6600", "matching_algorithm": 3, "match": "2025 W2,2025 1099,tax 2025"}'
curl -s -X POST -H "$AUTH" \
"$PAPERLESS/documents/post_document/" \
-F "document=@/path/to/file.pdf" \
-F "title=Tax Return 2025" \
-F "tags=5,8"
1 — Any word2 — All words3 — Exact match4 — Regular expression5 — Fuzzy match6 — Auto (ML-based)finance/tax-prep)?tags__id__isnull=true&ordering=-addedtax-2024, tax-2025receipt, invoice, statement, contract, medical, insuranceneeds-review, filed, archivedderek, shared, businessAPI returns 25 results by default. Use ?page=2 or ?page_size=100 (max 100).
query= for full-text search across OCR contentordering=-created for newest first, ordering=created for oldestcontent field in document details contains the full OCR text