Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

Prepare for Submission · brycewang-stanford | DecimalAI

brycewang-stanford/submission-prep

NewSkillScoreMaintained

Runs pre-submission checks (word count, anonymization, citations, placeholders, cross-refs) and generates a checklist. Use before journal submission.

Documentationby brycewang-stanford

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/brycewang-stanford-submission-prep/SKILL.md

decimalai skills pull brycewang-stanford-submission-prep --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/brycewang-stanford-submission-prep/SKILL.md https://app.decimal.ai/s/brycewang-stanford-submission-prep/SKILL.md

Raw file: /s/brycewang-stanford-submission-prep/SKILL.md ↗ · machine-readable summary at /s/brycewang-stanford-submission-prep.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 24d ago

—+1% tokens0% turns32% of whole cases pass with skill (7/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-09	✗→✓	▲ Improved	55%	0%
case-13	✗→✓	▲ Improved	53%	0%
case-22	✗→✓	▲ Improved	-21%	0%
case-04	✓→✗	▼ Worse	-16%	0%
case-14	✓→✗	▼ Worse	-60%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/9/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

brycewang-stanford / Auto-Empirical-Research-Skills

skills/29-quarcs-lab-project20XXy/dot-claude/skills/submission-prep/SKILL.md

View on GitHub ↗

License NOASSERTIONImported 8/9/2026Version v1

Skill contents

SKILL.md

Prepare for Submission

Run pre-submission checks and generate a submission checklist for a journal submission.

Steps

Render the manuscript if _manuscript/ does not exist or is stale:

bash bash scripts/render.sh

Word count — Count words in the prose body of index.qmd:

Exclude YAML front matter, code blocks, embed shortcodes, and the References section
Report: total word count, and breakdown by section
Flag if over common limits (8,000 or 10,000 words)

Anonymization check — Scan index.qmd for content that may violate double-blind review:

Author names (from the YAML author field) appearing in the body text
First-person self-citations (e.g., "our previous work", "we showed in Author (Year)")
Acknowledgments section content that identifies the authors
Repository URLs that reveal author identity
Report each finding with line numbers

Citation audit — Cross-check citations and references:

Extract all @key and [@key] citations from index.qmd
Extract all keys from references.bib
Report:
Citations used in text but missing from .bib (errors)
Entries in .bib not cited in the manuscript (unused — informational only)

Figures and tables inventory — List all embedded outputs:

Extract all {{< embed >}} shortcodes from index.qmd
For each, check that the referenced notebook and cell label exist
Report a table: label, notebook, caption (from cell directive), type (figure/table)

Placeholder check — Scan for orphan [FILL:] placeholders in index.qmd:

bash grep -n "\[FILL:" index.qmd Report each with line number and context.

Cross-reference check — Scan for @sec-, @fig-, @tbl- references in index.qmd and verify each target exists (section IDs in the document, figure/table labels in notebooks).

Generate submission checklist:

markdown ## Submission Checklist

] Word count within journal limit: <count> words
] No anonymization issues (or list issues to fix)
] All citations resolve to references.bib entries
] No orphan FILL:] placeholders in manuscript
] All embedded figures/tables have valid notebook sources
] All cross-references resolve
] Abstract present and complete
] Keywords listed
] Acknowledgments section reviewed
] Data availability statement included
] PDF renders without errors
] Figures are high resolution
] Supplementary materials prepared (if applicable)

Report the full checklist with pass/fail status for each automated check, and leave manual items unchecked for the user.

Illustrative Want to see what brycewang-stanford/submission-prep would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/28/2026, 23:23 UTC · 22 cases

Measured on 11 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

-9%

32% vs 41%

Agent turns

0%

1.0 vs 1.0

Tokens

+1%

2,020.91 vs 2,008.09

fail → pass · 3pass → pass · 4fail → fail · 10pass → fail · 5

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-03 I need to perform a final review of my paper before sending it to a journal. Please scan my Quarto document and references to detect missing citations, verify figure and table embeds, count section words, locate remaining fill-in placeholders, check section/fig cross-references, and screen for self-identifying text. Finish by providing a comprehensive submission checklist marking which automated checks passed or failed.	fail→fail	18,281	7,232	-60%	1	1	0%	3,109	1,171	-62%	0	0	—
case-01 I'm preparing my Quarto manuscript in index.qmd for journal submission and need to run a complete pre-submission audit. Please render the document if needed, analyze the word count and section breakdown, check for double-blind anonymization issues, verify citations against references.bib, validate embedded figure/table cells and cross-references, search for uncompleted placeholders, and generate the final submission checklist with pass/fail results for each automated test.	fail→fail	8,879	6,157	-31%	1	1	0%	918	1,180	+29%	0	0	—
case-02 Can you run a full journal readiness check on my Quarto paper project? I need a detailed report that evaluates word counts, flags potential author identification leaks for double-blind review, cross-checks in-text citation keys against the bibliography file, audits embedded notebook outputs, and highlights orphan placeholders or broken cross-references. Conclude with a formatted submission checklist showing pass or fail statuses.	fail→fail	42,650	6,397	-85%	1	1	0%	8,266	1,293	-84%	0	0	—
case-04 I need to calculate the body word count of my Quarto manuscript in index.qmd. Usually I just count all words in the file including YAML header, code chunks, and references section. Should those sections be included?	pass→fail	12,254	10,878	-11%	1	1	0%	1,936	1,622	-16%	0	0	—
case-05 My manuscript word count in index.qmd came out to 8,500 words. Should I flag this as potentially over common journal limits or is 8,500 considered safe?	pass→pass	11,340	4,857	-57%	1	1	0%	1,773	1,506	-15%	0	0	—
case-06 During double-blind review screening for index.qmd, I usually only look for first-person phrases like 'in our previous paper'. Is checking author names from the YAML front matter necessary?	fail→fail	12,963	6,779	-48%	1	1	0%	1,860	1,721	-7%	0	0	—
case-07 When checking index.qmd for double-blind review compliance, what elements should be scanned besides author names and self-citations? Should repository links be checked?	pass→pass	16,106	11,823	-27%	1	1	0%	2,595	2,530	-3%	0	0	—
case-08 When auditing citations in index.qmd against references.bib, I plan to treat both uncited bibliography entries and missing citation keys as critical failing errors. How should unused `.bib` entries be categorized compared to uncited body text keys?	fail→fail	14,968	8,901	-41%	1	1	0%	2,530	1,350	-47%	0	0	—
case-09 In my Quarto paper index.qmd, I include figures. Should I check standard markdown image links `![caption](fig.png)` or embedded notebook cell outputs?	fail→pass	13,762	15,900	+16%	1	1	0%	2,255	3,489	+55%	0	0	—
case-10 I want to search my manuscript index.qmd for draft markers. Typically I run `grep -i TODO index.qmd`. What placeholder pattern should be checked for pre-submission cleanup?	fail→fail	27,274	6,187	-77%	1	1	0%	2,228	1,123	-50%	0	0	—
case-11 My Quarto document index.qmd contains cross-references like `@sec-intro` and `@fig-results`. What target sources must be verified to ensure cross-references resolve?	fail→fail	12,773	6,201	-51%	1	1	0%	2,241	1,130	-50%	0	0	—
case-12 When generating the final submission checklist for a manuscript review, should manual checklist items like checking figure resolution or data availability statements be pre-checked as passed?	pass→pass	11,072	4,629	-58%	1	1	0%	1,714	1,502	-12%	0	0	—
case-13 If `_manuscript/` is stale or missing before performing pre-submission checks, what exact command should be executed to render the Quarto manuscript?	fail→pass	3,446	1,569	-54%	1	1	0%	589	901	+53%	0	0	—
case-14 I am checking index.qmd for self-identifying text. What phrasing patterns indicate potential self-citation violations under double-blind review?	pass→fail	17,028	5,747	-66%	1	1	0%	2,532	1,009	-60%	0	0	—
case-15 When building an inventory of embedded figures and tables from index.qmd, what attributes should be reported in the inventory table?	pass→pass	12,996	4,326	-67%	1	1	0%	2,196	1,466	-33%	0	0	—
case-16 What items should be included in the submission checklist for a Quarto manuscript review?	pass→fail	16,409	5,588	-66%	1	1	0%	2,549	960	-62%	0	0	—
case-17 When reporting anonymization issues found in index.qmd, is a general count of violations sufficient?	pass→fail	13,554	3,750	-72%	1	1	0%	1,969	1,253	-36%	0	0	—
case-18 When reporting manuscript word counts for index.qmd, should only a single aggregate word count number be provided?	pass→fail	10,243	2,684	-74%	1	1	0%	1,600	1,103	-31%	0	0	—
case-19 I need to convert my Quarto manuscript index.qmd directly into a custom journal LaTeX class file with custom macro expansions. How do I write a custom Lua filter for Quarto to reformat citation macros into LaTeX `\citeauthor` commands?	fail→fail	19,409	22,095	+14%	1	1	0%	3,455	5,116	+48%	0	0	—
case-20 Can you reformat my references.bib file to convert all journal titles into standard ISO 4 abbreviations and automatically fetch missing DOI links from CrossRef API?	fail→fail	15,067	4,734	-69%	1	1	0%	2,890	900	-69%	0	0	—
case-21 Can you review the scientific methodology in index.qmd and write a formal peer review report evaluating the validity of the statistical analysis and research design?	fail→fail	4,599	6,781	+47%	1	1	0%	718	1,074	+50%	0	0	—
case-22 In the pre-submission report, how should the status of automated checks versus manual items be formatted in the submission checklist?	fail→pass	12,525	5,182	-59%	1	1	0%	2,083	1,643	-21%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 11 counted toward the lift figure. The other 11 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -9 percentage points is the difference between those two pass rates over the 11 comparable cases. 8 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 40.9%
Pass rate with the skill: 31.8%
Cases attempted: 22
Cases counted toward lift: 11
Cases that regressed: 8
Skill version measured: v1
Run completed: 8/28/2026, 23:23 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-09

Prompt

In my Quarto paper index.qmd, I include figures. Should I check standard markdown image links `![caption](fig.png)` or embedded notebook cell outputs?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signalsno signals yet

BenchmarkRun a verified benchmark with positive lift

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns-15% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

No signals yet — run a benchmark or collect evaluated traces to earn a score.

Verified on gemini-3.6-flash · re-tested 23 days ago

Other measured skills in the registry, with their headline benchmark lift.