Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/capitalparser/notebooklm-wiki-pipelinenpx agentmods add commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wikiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wiki)<a href="https://agentmods.dev/commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wiki"><img src="https://agentmods.dev/badge/commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wiki/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wiki"><img src="https://agentmods.dev/badge/commands/capitalparser/notebooklm-wiki-pipeline/pdf-to-wiki.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00000 | $0.02496 |
| Opus 5 | $0.00000 | $0.01248 |
| Sonnet 5 | $0.00000 | $0.00499 |
| Haiku 4.5 | $0.00000 | $0.00250 |
Grade A, and why
pdf-to-wiki scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 245 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/pdf-to-wiki
Drive에 있는 대용량 PDF를 NotebookLM에 외주화하여 분석하고,
결과를 OUTPUT_DIR에 Obsidian 노트로 저장한다.
기본 출력 경로:
OUTPUT_DIR=~/vault/00_Wiki/AI_Generated
설치 후 위 경로를 본인 Obsidian vault의 저장 위치로 바꿔라.
중요: PDF 파일 내용을 직접 읽지 않는다. Drive URL만 NotebookLM에 전달한다.
mcp__claude_ai_Google_Drive__download_file_content 호출 금지.
사용법
/pdf-to-wiki <drive_url_or_file_id> [wiki_title] [--topic topic_id]
drive_url_or_file_id: Google Drive 파일 URL 또는 파일 IDwiki_title(선택): 생성할 Wiki 노트 제목. 생략 시 파일명에서 추론.--topic(선택):config/notebooks.example.json형식의 registry에 있는 주제 ID. 지정하면 해당 NotebookLM 노트북을 재사용한다.
Notebook 재사용 정책
기본 정책은 여전히 안전한 single_source_notebook이다. 주제가 불명확하거나 registry에 없는 자료는 새 NotebookLM 노트북을 만든다.
주제별 Notebook 재사용은 아래 조건에서만 수행한다.
- 사용자가
--topic을 명시했거나, 제목이 registry의routing_keywords와 매칭됨 - 해당 topic의 NotebookLM 노트북이 문서 세트 관점에서 좁게 정의되어 있음
- 같은 Drive file ID가 이미
sources에 있으면 source를 다시 추가하지 않고 기존 source를 재사용함
재사용의 이점:
- 같은 주제의 기존 PDF와 신규 PDF를 한 노트북에서 비교할 수 있음
- 기존 기준서, 리포트, 메모와 연결되는 개념을 NotebookLM이 더 잘 찾을 수 있음
- topic-level 질문에서는 누적 문서 세트 전체를 대상으로 답변을 받을 수 있음
단, 신규 PDF 노트를 만들 때는 답변 오염을 막아야 한다. 같은 NotebookLM 노트북에 있는 이전 PDF까지 섞여 답변될 수 있으므로, 기본 notebook_query는 MCP 인자의 source_ids=[target_source_id]로 대상 PDF source만 지정한다. 프롬프트에서도 대상 PDF만 primary scope로 제한하고, 기존 PDF는 비교/연결 섹션에서만 참고하게 한다.
기본 추출 모드는 source_scoped_topic_query다.
- 대상 PDF를 topic notebook에 추가하거나 이미 등록된 source를 찾는다.
- 신규 Wiki 노트용
notebook_query는 topic notebook에서 실행하되 MCP 인자source_ids=[target_source_id]를 전달하고, 프롬프트에도 targetsource_id또는drive_file_id를 명시한다. - 기존 topic source는 비교/연결 섹션에서만 사용한다.
- MCP schema가 특정 source 지정 query를 지원하지 않거나 불명확할 때만
single_source_first를 fallback으로 사용한다.
라우팅 결정은 로컬 helper로 먼저 확인할 수 있다.
python3 scripts/notebook_registry.py \
"https://drive.google.com/file/d/YOUR_FILE_ID/view" \
--title "K-IFRS 1109 금융상품" \
--topic audit-accounting \
--registry config/notebooks.local.json
실행 절차
아래 단계를 순서대로 실행한다. notebooklm-mcp 서버가 활성화되어 있어야 한다.
인증 오류 발생 시: nlm login 실행 후 재시도.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 245 lines · 0 tokens per session scan A 6b4df8bce789
pdf-to-wiki is a command published in the GitHub repository capitalparser/notebooklm-wiki-pipeline (93 stars, last pushed 3mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,496 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
wiki-lint
Run a health check on the wiki. Invoke with /wiki-lint or "lint the wiki".
okf
A command that exports a private knowledge wiki into an OKF-compatible bundle. OKF is a format for packaging knowledge, with a separate guarded mode for preparing material to share externally.
ingest
Compile new sources from raw/ into the wiki.
pdf-to-markdown-docling
Convert a local PDF to markdown via Docling's standard pipeline (layout + table-structure recognition — higher fidelity than MarkItDown on complex tables/layouts, 10x slower). Requires the opt-in Docling install.
pdf-to-markdown
Convert a local PDF to markdown via the bundled MarkItDown Python CLI (fast, lightweight — plain text extraction, no table-structure recognition). For complex tables/layouts prefer /obsidian-router:pdf-to-markdown-docling.
ingest
Ingest a source (URL/file/text) into Raw Sources + compile 1015 Wiki pages, with mandatory user-purpose gate and mothership cross-linking.