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 clearsmog/claude-skills --skill create-documentgit clone --depth 1 https://github.com/clearsmog/claude-skillsWrote 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/clearsmog/claude-skills/create-document)<a href="https://agentmods.dev/skills/clearsmog/claude-skills/create-document"><img src="https://agentmods.dev/badge/skills/clearsmog/claude-skills/create-document/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/clearsmog/claude-skills/create-document"><img src="https://agentmods.dev/badge/skills/clearsmog/claude-skills/create-document.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 | $0.00032 | $0.01785 |
| Opus 5.5 | $0.00013 | $0.00714 |
| Sonnet 5.5 | $0.00006 | $0.00357 |
| Haiku 4.5 | $0.00003 | $0.00178 |
Grade A, and why
create-document 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 24d 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 — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Document Creation Workflow
One prompt → one publication-quality document. Internal quality loop runs automatically.
Format Selection
- User specifies format explicitly → use that
- User doesn't specify → detect from project context:
- Has
Slides/directory → Beamer LaTeX - Has
Source/or mainly.typfiles → Typst - Has
Quarto/directory → Quarto
- Has
- Default for new projects → Typst (preferred modern format)
Format Detection (for .typ)
Detect document type from user request keywords:
- resume, CV → CV template
- slides, presentation, lecture → touying slides (see
typst/references/touying-guide.md) - essay, paper → essay template
- guide, study guide, reference → study guide
- report, business → business report
- reference card, cheat sheet → dense reference layout
Constraints
All formats: verify every citation against the bibliography.
Teaching material (lectures, slides, study guides): first read the project's knowledge base in .claude/rules/ (notation registry, narrative arc) and check each new symbol against it; motivation before formalism; a worked example within 2 slides/pages of each definition; transition slides at major conceptual pivots.
WHEN TO PAUSE (the ONLY case)
Request is genuinely ambiguous AND no project context to disambiguate:
- No sibling
.typfiles, no PDFs, no clear topic in the prompt - Example: user says "make something" with zero context
In this case: ONE AskUserQuestion (topic + type), then proceed autonomously.
For everything else — proceed without asking. Template auto-selection is correct 90%+ of the time; user can re-invoke with explicit overrides if wrong.
PIPELINE
Phase 0: Auto-Discover
Run these before drafting, without stopping to confirm:
- Scan project for related materials
- Glob for
*.pdf,*.md,*.typ,images/in the project directory - If PDFs found → check for
.parsed.mdnext to each; read if exists - If no
.parsed.md→ parse using PDF handling rules (background for large files)
- Glob for
- Detect document type from keywords + project context (see Format Detection above)
- Auto-select template — no confirmation needed
- Style inheritance: if sibling
.typfiles exist in the project:- Read first 30 lines of the most recent sibling
- Match fonts, colors, qk preset
- Only inherit from files that use
@local/qkpresets; ignore raw inline styles
- Build context: materials found, template chosen, style detected
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.
- 24d ago Changed · -27 lines 4398a3f1b187
- 1mo ago Changed · +1 lines ce20b973babc
- 1mo ago First seen · 214 lines · 32 tokens per session scan A 4e5873a3dc2f
create-document is a skill published in the GitHub repository clearsmog/claude-skills (2 stars, last pushed 24d ago), licensed MIT. It adds 32 tokens to every session and 1,785 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 2026-08-31.
Other skills, from other repositories
surfsense-documents
Make or change a Word document (.docx), a PDF, a PowerPoint deck (.pptx) or an Excel workbook (.xlsx) by writing a Python script that surfsenserenderdocument runs, optionally starting from a source's own Word or PowerPoint file. Load it whenever the user asks for one of these, or for a change to one you made.
surfsense-html
Make or change a web page from the user's sources - a report, a briefing, a dashboard, a comparison, a timeline, an explainer, an interactive calculator or a figure that shows how something works - by writing an HTML file and keeping it with surfsensesavepage. Load it whenever the user asks for a web page, an HTML…
surfsense-data
Work out numbers from spreadsheet and CSV sources (.xlsx, .csv, .tsv) by writing a pandas script that surfsenseanalyzedata runs: totals, averages, trends, rankings, comparisons, cleaning, and charts that can go into a document. Load it whenever the user asks for figures from a table, or a chart of one.
surfsense-pdf
Work on PDFs the user has, or PDFs in Studio, without changing them - merge, take out, split, turn or reorder pages, add a watermark, page numbers, a header or a footer, and list or fill a form's fields. Load it whenever the user asks for one of these on a PDF.
pptx
Create polished, editable PowerPoint decks for slide presentations.
xlsx
Create polished Excel workbooks for budgets, trackers, and data tables.