agentleFS
Sign inStart free
boettiger-labboettiger-lab/agent-skills/reproducible-reports/SKILL.md

reproducible-reports

Skill4 starsChanged 16 days agoBSD-3-ClauseView on GitHub

How to deliver reports, write-ups, analyses and replications for Carl, using an open, reproducible academic stack (pandoc + LaTeX PDF, markdown source, scripts and SQL in the repo, Makefile) rather than Claude Docs, Word or other proprietary or hosted formats. Use whenever asked to "write up", "make a report", or replicate or summarise an analysis.

  • Installs packages

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Prose: follow the writing-voice skill: plain headings, declarative sentences, and caveats stated as facts.

    line 17
  2. Very specific:

    Do not credit Claude as an author.

    line 36

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Reproducible reports

Carl wants reports he can rebuild, version and share as an academic would. Do not make a Claude Doc, a .docx or a hosted page for this. Use those only if he asks for one by name.

Default stack

  • Source: report.md (pandoc markdown) plus references.bib, cited with --citeproc.
  • Output: a PDF article via LaTeX (pandoc --pdf-engine=xelatex). Fall back to stand-alone HTML (pandoc -s --embed-resources) only when LaTeX truly cannot be installed.
  • Code: the queries (queries/*.sql) and scripts that produce every number, table and figure live next to the report. Do not type numbers into the prose by hand. Generate them (for example, build/values.json substituted into {{placeholders}}) so a rerun updates the text.
  • Build: a Makefile that goes results → analysis → report.pdf. Keep the slow data step as its own target so results can be cached.
  • Figures: matplotlib or ggplot written to files, 300 dpi PNG or vector PDF.
  • Prose: follow the writing-voice skill: plain headings, declarative sentences, and caveats stated as facts.

Tools are installable

The pods have passwordless sudo apt-get. If pandoc, LaTeX or a font is missing, install it rather than downgrading the output. A typical install:

sudo DEBIAN_FRONTEND=noninteractive apt-get install -y texlive-xetex texlive-latex-recommended \
  texlive-latex-extra texlive-fonts-recommended lmodern

Quarto is also available (/usr/lib/rstudio-server/bin/quarto) if executable chunks are a better fit. Plain pandoc is the default.

Data from the MCP servers

To keep the analysis reproducible outside a chat session, call the MCP server from a script rather than pasting tool results into the report. Use JSON-RPC over HTTP: initialize, then tools/call with query. The server returns markdown tables previewed at about 50 rows. For larger results, pack them into one cell (string_agg(concat_ws(',', ...), ';')) and unpack them in the script. A working client is at ~/boettiger-lab/oneill2026-replication/mcp_query.py.

Disclosure

Do not credit Claude as an author. Follow the disclose-tool-use skill.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free