Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Mexregkan/claude-for-researchers --skill wolfbookgit clone --depth 1 https://github.com/Mexregkan/claude-for-researchersWrote 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/mexregkan/claude-for-researchers/wolfbook)<a href="https://agentmods.dev/skills/mexregkan/claude-for-researchers/wolfbook"><img src="https://agentmods.dev/badge/skills/mexregkan/claude-for-researchers/wolfbook/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/mexregkan/claude-for-researchers/wolfbook"><img src="https://agentmods.dev/badge/skills/mexregkan/claude-for-researchers/wolfbook.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Anti-Refusal · line 124 Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.Fix: Remove instructions that suppress warnings, disclaimers, or ethical commentary. Let the agent surface safety-relevant caveats to the user.
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 | $0.02683 |
| Opus 5 | $0.00000 | $0.01341 |
| Sonnet 5 | $0.00000 | $0.00537 |
| Haiku 4.5 | $0.00000 | $0.00268 |
Grade A, and why
wolfbook 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 12d 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 — 140 lines — stays where its author put it; the contents beside it link to each section on GitHub.
wolfbook — driving the live Wolfram kernel + notebook via MCP
The mcp__wolfbook__* tools talk to a running Wolfram kernel inside a VS Code window, editing
and evaluating a live .wb/.nb notebook. This is the opposite end from wolfram-headless
(one-shot wolframscript): here state persists, cells have stable IDs, and you can
checkpoint/restore.
Requires the Wolfbook extension with the MCP enabled
("wolfbook.mcpEnabled": true); without it the mcp__wolfbook__* tools are not available — use
wolfram-headless for .wls runs instead. Tool schemas are deferred — fetch them with
ToolSearch "select:mcp__wolfbook__<name>" before first use.
Decision: MCP live kernel vs headless wolframscript
- Live kernel (MCP) — interactive work on the project notebook: read/edit/run cells, probe
state, small-to-medium evaluations, debugging, building up definitions incrementally. State is
preserved between calls;
kernelControlgives real checkpoint/restore. - Headless (
wolframscript, seewolfram-headless) — heavy batch jobs (a big symbolic solve) that would block or OOM the interactive kernel, or that you want to run in the background and poll a result file. The live kernel in VS Code contends for the license seat with headless jobs — close/idle the Wolfbook window before a long headless run, or target a worker client. - Both report a kernel crash as "The product exited because of a license error" — it is almost
never licensing (see
wolfram-headlessRULE 2). On the MCP side usewolfbook_kernelCrashLog(source:"debug"/"crash") to get the real stack/cause.
Tool map (right tool for the job)
Orient before touching anything
wolfbook_list_clients— list VS Code windows, their open notebooks, primary/worker role.wolfbook_setTarget {client_id, notebook}— pin a default so you stop passing them every call (set it once at session start and omitnotebookfrom subsequent calls).wolfbook_getNotebookContext action:"read"(orbrief:true/action:"summary") — full or one-line-per-cell view. Always read first; cell numbers shift after structural edits, so grab the stable CellId and use that. (Its outputs are a CACHED snapshot — see "Read kernel errors" below.)wolfbook_searchCells {query, regex}— locate a symbol/def/error in a long notebook without reading it whole; returns CellId + whether the hit was in source or output.wolfbook_getKernelState {pattern}— list definedGlobal*` symbols (values / DownValue counts) before editing, to avoid clobbering an existing definition (watch for name-collision hazards; record known ones in your CLAUDE.md).wolfbook_getCellOutput {cellId}— read a cell's existing (stored) output without re-running it.
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.
- 12d ago First seen · 140 lines · 0 tokens per session scan A 70d4756c1bba
wolfbook is a skill published in the GitHub repository Mexregkan/claude-for-researchers (52 stars, last pushed 8d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,683 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 skills, from other repositories
pre-submission-report
Run the final, comprehensive submission-readiness gate and consolidate all checks into one dated report; citation-integrity-only mode is also supported. Use when a paper and submission package are nearly final. Not for a mid-draft adversarial review; use $review-cluster.
proof-readability
Improve the exposition and readability of a mathematical proof already verified as correct without changing its mathematics. Use when polishing a lemma, theorem, proof, or appendix after correctness checks. Not for verifying the proof; use $verify-math.
latex
Compile one specified LaTeX document, resolve build errors, audit citations, and report build quality. Use when a .tex source must be built or a concrete compilation failure repaired. Not for corpus-wide build checks, visual polish after a clean build, or creating a project; use $latex-health-check, $latex-polish, or…
beamer-deck
Create an academic presentation as a LaTeX Beamer source and reviewed PDF with an original theme. Use when the requested deliverable is a conference, seminar, or lecture deck in Beamer. Not for PowerPoint or RevealJS; use $pptx or $quarto-deck.
bib-parse
Extract citations from a PDF and generate a validated .bib file. Use when the user asks to extract citations from a PDF and generate a validated .bib file. Reads the PDF, identifies referenced works, constructs BibTeX entries, and verifies metadata.
computational-experiments
Scaffold, execute, analyse, and publish computational research experiments through a reproducible staged workflow. Use when a research question requires simulations or computational sweeps rather than a one-off script.