Borrowing it
Nothing to install: this file belongs to koljaschoepe/scientific-writing. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/koljaschoepe/scientific-writing/main/.claude/skills/recherche/SKILL.mdgit clone --depth 1 https://github.com/koljaschoepe/scientific-writingWrote 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/skills/koljaschoepe/scientific-writing/recherche)<a href="https://agentmods.dev/skills/koljaschoepe/scientific-writing/recherche"><img src="https://agentmods.dev/badge/skills/koljaschoepe/scientific-writing/recherche/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/skills/koljaschoepe/scientific-writing/recherche"><img src="https://agentmods.dev/badge/skills/koljaschoepe/scientific-writing/recherche.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 | $0.00022 | $0.01056 |
| Opus 5.5 | $0.00009 | $0.00422 |
| Sonnet 5.5 | $0.00004 | $0.00211 |
| Haiku 4.5 | $0.00002 | $0.00106 |
Grade A, and why
recherche 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 26d 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 — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/recherche: Literatur finden
Wichtig: Nie eine Quelle, DOI oder Jahreszahl erfinden, jede DOI gegen Crossref prüfen.
Claude legt nur Vorschläge an, die Person entscheidet. .arbeit/kandidaten.json nie lesen
oder von Hand schreiben, nur node .claude/kit/werkzeuge/kandidaten.mjs. Endausgabe laut AGENTS.md.
Argument: $ARGUMENTS (z. B. 2.3 für Kapitel 2.3 oder "yield prediction")
1. Auftrag klären (höchstens eine Interview-Runde)
Nur wenn das Argument nicht eindeutig ist und „So arbeite ich“ in .arbeit/stil.md nichts sagt:
- „Wofür suchen wir?“ header „Ziel“: aus Thema bzw. Gliederung, z. B. „Stand der Forschung zu (Empfohlen)“, „Methoden für “, „Belege für Kapitel “, „Gegenpositionen“.
- „Wie breit?“ header „Umfang“: „10 bis 15 starke Treffer (Empfohlen)“, „Breiter Überblick (30+)“, „Nur die 5 wichtigsten“.
- „Wo suchen?“ header „Quellen“, multiSelect: „Fachdatenbanken (Empfohlen)“, „Bibliothek und Verlage mit Uni-Login“, „Google Scholar“, „Allgemeine Websuche“ (Browser dauert länger).
2. Suchen
- Fachdatenbanken: Subagent
rechercheur. Auftrag: Ziel, Kapitel, Anzahl, Zeitraum, bekannte DOIs zum Ausschließen. Er trägt selbst ins Board ein und gibt nur eine kurze Zusammenfassung zurück. Beitechnik.abo: proein Rechercheur, nicht mehrere parallel. - Browser (Playwright) im Hauptgespräch, wenn gewählt: Katalog aus
hochschule.bibliothek(leer: Bibliotheksseite der Hochschule per Websuche finden, bestätigen lassen, eintragen). Beim ersten Mal die Anmeldeseite öffnen und per Interview bitten, sich selbst anzumelden („Bin angemeldet“, „Klappt nicht“, „Überspringen“). Langsam und einzeln. Jede DOI prüfen. - Websuche nur für Hintergrund, Software-Doku, graue Literatur.
Funde aus dem Browser selbst eintragen, alle auf einmal:
node .claude/kit/werkzeuge/kandidaten.mjs add - <<'JSON'
{"quellen": [{"id": "<doi oder url>", "titel": "", "autoren": ["Nachname, Vorname"],
"jahr": 2024, "venue": "", "doi": "", "url": "", "kurz": "2 Sätze Inhalt",
"warum": "1 Satz ohne Jargon", "kapitel": ["2.1"], "relevanz": 4,
"markierungen": ["review"], "open_access": null, "herkunft": "bibliothek"}]}
JSON
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.
- 26d ago Changed · +5 lines · -2 tokens per session cfb225679d1c
- 28d ago First seen · 67 lines · 24 tokens per session scan A 10f353375a82
recherche is a skill published in the GitHub repository koljaschoepe/scientific-writing (13 stars, last pushed 26d ago), licensed MIT. It adds 22 tokens to every session and 1,056 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-10-01.
Other skills, from other repositories
ml-paper-writing
Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. Includes LaTeX templates, reviewer guidelines, and citation verification workflows.
academic-plotting
Generates publication-quality figures for ML papers from research context. Given a paper section or description, extracts system components and relationships to generate architecture diagrams via Gemini. Given experiment results or data, auto-selects chart type and generates data-driven figures via matplotlib/seaborn.…
systems-paper-writing
Comprehensive guide for writing systems papers targeting OSDI, SOSP, ASPLOS, NSDI, and EuroSys. Provides paragraph-level structural blueprints, writing patterns, venue-specific checklists, reviewer guidelines, LaTeX templates, and conference deadlines. Use this skill for all systems conference paper writing.
overleaf
Sync and manage Overleaf LaTeX projects from the command line. Pull projects locally, push changes back, compile PDFs, and download compile outputs like .bbl files for arXiv submissions. Use when working with LaTeX, Overleaf, academic papers, or arXiv.
paper-skill
End-to-end, Chinese-first academic research and publication workflow for reading papers, literature reviews, research framing, manuscript planning and writing, the eight-step paper-system workflow, Chinese-English translation, LaTeX, figures, citation verification, reporting-guideline checks, journal selection, peer…
parallel-subagents
When a task decomposes into independent slices — separate files, OR separate sections of ONE file — fan it out: write job files to the subagent queue and the host runs real parallel Codex workers for you. For a single file, extract per-section slice files into the queue's work/ zone, let workers polish them in…