Ready to supercharge your AI agents?

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

Get Started Free →

Paper Digest · paperclipai | DecimalAI

paperclipai/paper-digest

NewSkillScoreMaintained

Scan and summarize recent academic papers from Semantic Scholar

researchby paperclipai

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/paperclipai-paper-digest/SKILL.md

decimalai skills pull paperclipai-paper-digest --out .claude/skills/

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

curl -fsSL --create-dirs -o .claude/skills/paperclipai-paper-digest/SKILL.md https://app.decimal.ai/s/paperclipai-paper-digest/SKILL.md

Raw file: /s/paperclipai-paper-digest/SKILL.md ↗ · machine-readable summary at /s/paperclipai-paper-digest.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 54d ago

—+0% tokens0% turns14% of whole cases pass with skill (3/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-20	✓→✓	＝ Same ✓	-33%	0%
case-21	✓→✓	＝ Same ✓	23%	0%
case-22	✓→✓	＝ Same ✓	34%	0%
case-01	✗→✗	＝ Same ✗	-6%	0%
case-02	✗→✗	＝ Same ✗	43%	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 — 8/1/2026

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

paperclipai / companies

aeon-intelligence/skills/paper-digest/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v1

Skill contents

SKILL.md

Scans Semantic Scholar for recent papers matching tracked topics and produces curated summaries.

Illustrative Want to see what paperclipai/paper-digest 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

8/4/2026, 16:54 UTC · 22 cases

Measured on 21 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

0%

14% vs 14%

Agent turns

0%

1.0 vs 1.0

Tokens

+0%

2,489.43 vs 2,481.1

pass → pass · 3fail → fail · 19

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-01 Could you search Semantic Scholar for the latest research papers on transformer architecture optimizations and pull together a brief digest of the main findings?	fail→fail	13,263	18,185	+37%	1	1	0%	2,177	2,045	-6%	0	0	—
case-02 I'm tracking recent publications about reinforcement learning from human feedback. Please check for newly released papers on this topic and provide a curated summary of each.	fail→fail	14,986	19,503	+30%	1	1	0%	2,447	3,499	+43%	0	0	—
case-03 Can you scan academic literature for recent papers covering graph neural networks in drug discovery and compile a structured summary highlighting their core methodologies?	fail→fail	20,859	17,361	-17%	1	1	0%	3,701	3,198	-14%	0	0	—
case-04 I want to keep tabs on recent developments in diffusion models applied to medical image segmentation. Can you check Semantic Scholar for recent papers on this topic and summarize their main contributions?	fail→fail	13,134	15,638	+19%	1	1	0%	2,112	2,599	+23%	0	0	—
case-05 Find recent literature on surface codes and fault-tolerant quantum error correction published recently, and give me a brief summary digest of what's new.	fail→fail	15,893	6,825	-57%	1	1	0%	2,547	285	-89%	0	0	—
case-06 Please scan Semantic Scholar for recent research on evaluating Retrieval-Augmented Generation (RAG) systems and provide a digest summarizing the proposed metrics.	fail→fail	15,034	18,437	+23%	1	1	0%	2,622	3,264	+24%	0	0	—
case-07 Can you search Semantic Scholar for recent work on LoRA variants and parameter-efficient fine-tuning for large language models and produce a concise summary digest?	fail→fail	14,681	12,269	-16%	1	1	0%	2,673	2,318	-13%	0	0	—
case-08 I am following research on machine learning models predicting climate change impacts on agricultural crop yields. Search Semantic Scholar for recent papers and provide a summary of findings.	fail→fail	15,090	14,079	-7%	1	1	0%	2,358	1,919	-19%	0	0	—
case-09 Search Semantic Scholar for recent publications on spiking neural network hardware implementations and compile a summary digest of the recent advances.	fail→fail	17,199	19,069	+11%	1	1	0%	2,665	2,972	+12%	0	0	—
case-10 Find newly published papers on multimodal sentiment analysis combining audio, text, and video on Semantic Scholar, and summarize their core findings.	fail→fail	16,859	20,321	+21%	1	1	0%	2,980	2,725	-9%	0	0	—
case-11 Scan Semantic Scholar for recent research on vision-language-action models for robotic manipulation and produce a brief summary digest.	fail→fail	13,351	11,717	-12%	1	1	0%	2,280	1,895	-17%	0	0	—
case-12 Check Semantic Scholar for recent papers using synthetic data generation to train clinical natural language processing models, and give me a curated summary.	fail→fail	15,634	19,696	+26%	1	1	0%	2,510	2,452	-2%	0	0	—
case-13 Search Semantic Scholar for recent literature on Direct Preference Optimization and its variants for alignment, and summarize the key results.	fail→fail	17,713	34,900	+97%	1	1	0%	3,165	2,715	-14%	0	0	—
case-14 Check Semantic Scholar for newly published papers on communication-efficient federated learning on edge devices and summarize their main contributions.	fail→fail	14,609	10,902	-25%	1	1	0%	2,381	1,709	-28%	0	0	—
case-15 Scan Semantic Scholar for recent papers on benchmarking large language models for code generation and software engineering tasks, and provide a digest.	fail→fail	17,957	22,077	+23%	1	1	0%	3,209	3,513	+9%	0	0	—
case-16 Search Semantic Scholar for recent publications on red-teaming and safety evaluation of autonomous AI agents, and summarize the primary findings.	fail→fail	14,963	19,589	+31%	1	1	0%	2,328	2,699	+16%	0	0	—
case-17 Check Semantic Scholar for recent research on scaling context windows in transformer models (e.g., ring attention, state space models) and produce a brief digest.	fail→fail	14,385	15,442	+7%	1	1	0%	2,088	2,665	+28%	0	0	—
case-18 Scan Semantic Scholar for recent papers applying pre-trained foundation models to zero-shot time series forecasting, and summarize their main techniques.	fail→fail	20,374	15,585	-24%	1	1	0%	2,923	2,788	-5%	0	0	—
case-19 Search Semantic Scholar for recent work on mechanistic interpretability and induction heads in language models, and compile a summary digest.	fail→fail	17,624	16,525	-6%	1	1	0%	2,923	2,653	-9%	0	0	—
case-20 I have three paper titles and authors I cited in my draft: 'Attention Is All You Need' by Vaswani et al., 'BERT' by Devlin et al., and 'GPT-3' by Brown et al. Please convert these exact references into properly formatted BibTeX entry blocks.	pass→pass	7,856	6,040	-23%	1	1	0%	1,943	1,300	-33%	0	0	—
case-21 I am writing a paper submission on efficient fine-tuning of multi-modal models. Here are my bullet points: 1) We introduce a low-rank projection adapter. 2) 4x reduction in GPU memory during training. 3) Tested on LLaVA benchmarks. Please format these points into a polished, 150-word academic abstract in LaTeX text.	pass→pass	8,190	8,187	-0%	1	1	0%	1,434	1,763	+23%	0	0	—
case-22 Here is the abstract text from a manuscript I received for peer review: 'Abstract: We present a novel sparse attention mechanism that reduces computational complexity from O(N^2) to O(N log N)...'. Please analyze this provided text and summarize its key methodology and limitations.	pass→pass	6,841	9,730	+42%	1	1	0%	1,184	1,587	+34%	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 21 counted toward the lift figure. The other 1 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 0 percentage points is the difference between those two pass rates over the 21 comparable cases. 1 case got worse with the skill loaded, and it is 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: 13.6%
Pass rate with the skill: 13.6%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 1
Skill version measured: v1
Run completed: 8/4/2026, 16:54 UTC

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+2% 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 54 days ago

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