cs.AIApr 5, 2026

SKILLFOUNDRY: Building Self-Evolving Agent Skill Libraries from Heterogeneous Scientific Resources

Authors: Shuaike Shen, Wenduo Cheng, Mingqian Ma, Alistair Turcan, Martin Jinye Zhang, Jian Ma

Organizations: Ray and Stephanie Lane Computational Biology Department, School of Computer Science, Carnegie Mellon University · Machine Learning Department, School of Computer Science, Carnegie Mellon University

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Modern scientific ecosystems are rich in procedural knowledge across repositories, APIs, scripts, notebooks, documentation, databases, and papers, yet much of this knowledge remains fragmented across heterogeneous artifacts that agents cannot readily operationalize. This gap between abundant scientific know-how and usable agent capabilities is a key bottleneck for building effective scientific agents. We present SkillFoundry, a self-evolving framework that converts such resources into validated agent skills, reusable packages that encode task scope, inputs and outputs, execution steps, environment assumptions, provenance, and tests. SkillFoundry organizes a target domain as a domain knowledge tree, mines resources from high-value branches, extracts operational contracts, compiles them into executable skill packages, and then iteratively expands, repairs, merges, or prunes the resulting library through a closed-loop validation process. SkillFoundry produces a substantially novel and internally valid skill library, with 71.1% of mined skills differing from existing skill libraries such as SkillHub and SkillSMP. We demonstrate that these mined skills improve coding agent performance on five of the six MoSciBench datasets. We further show that SkillFoundry can design new task-specific skills on demand for concrete scientific objectives, and that the resulting skills substantially improve performance on two challenging genomics tasks: cell type annotation and the scDRS workflow. Together, these results show that automatically mined skills improve agent performance on benchmarks and domain-specific tasks, expand coverage beyond hand-crafted skill libraries, and provide a practical foundation for more capable scientific agents.

Figures & tables

Figure 1: Overview of SkillFoundry . Starting from a domain knowledge tree, the framework mines branch-relevant resources, extracts structured candidate skills, validates them through multi-level testing, expands the tree with verified skills, and prunes redundant or low-value leaves.
Figure 2: Composition of the mined skill library and runtime of the mining pipeline. (a) Distribution of the 286 mined skills across domains and subdomains. (b) Average runtime of the major pipeline stages under a fixed resource-mining budget.
 health spamassspecgympop genetics
SettingRepo-Acc (%)Paper-Acc (%)Exec (%)Repo-Acc (%)Paper-Acc (%)Exec (%)Repo-Acc (%)Paper-Acc (%)Exec (%)
w/o skills64.7147.06100.0040.0040.00100.0053.8530.77100.00
w/ skills76.4758.82100.0046.6746.67100.0061.5446.15100.00
Table 1: Results on all six datasets in MoSciBench for the same coding agent with and without SkillFoundry skills. Repo-Acc uses the scoring rule implemented in the official MoSciBench repository, whereas Paper-Acc uses the scoring rule described in the MoSciBench paper. Exec denotes code execution success rate. SkillFoundry improves performance on five of the six datasets and leaves one unchanged, while maintaining perfect execution success throughout.
Figure 3: UMAP of cell-type annotations from ground truth, Codex (without skill), Codex+ SkillFoundry , and SpatialAgent. SMCs denote smooth muscle cells. Labels are harmonized into eight major cell types, following the SpatialAgent ( Wang et al., 2025a ) paper. Coverage is the fraction of cells assigned to one of the eight major cell types, and accuracy is the fraction whose predicted label matches the harmonized ground truth.
Figure 4: Quantitative and qualitative evaluation of Biomni on the scDRS workflow, with and without SkillFoundry skills. (a) RMSE between agent and expert outputs across three replicates. SkillFoundry reduces error overall, with two runs exactly matching the expert output. (b) Expert qualitative scores across the same replicates. A score of 7 indicates that all evaluation criteria are met. SkillFoundry produces the only run that satisfies all qualitative criteria.
Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

ReplicateSettingQualitative scoreRMSENotes
1Biomni30.16Filtering option not enabled
2Biomni40.16Minor deviation from expert pipeline
3Biomni40.00Exact scores but incomplete analysis
4Biomni + SkillFoundry60.06Nearly complete analysis
5Biomni + SkillFoundry70.00All criteria met, exact numerical match
6Biomni + SkillFoundry30.00Exact scores but incomplete reporting
Appendix
Table 2: Replicate-level scDRS evaluation. Replicates 1 to 3 correspond to Biomni without SkillFoundry skills, and replicates 4 to 6 correspond to Biomni with SkillFoundry skills.
Figure 5: Representative scDRS output generated by Biomni without SkillFoundry skills. The figure ranks cell types by association strength and marks nominal significance, but does not summarize FDR-supported signal or within-cell-type heterogeneity.
Figure 6: Representative scDRS output generated by Biomni with SkillFoundry skills. In addition to ranking cell types by association strength, the figure reports the proportion of FDR-significant cells within each cell type, providing a richer summary of statistical support and cell-type heterogeneity.
Figure 7: Web dashboard view of a cell type annotation skill. The interface shows searchable skill metadata, taxonomy placement, and links to associated artifacts and source resources.
Figure 8: Web dashboard view of scDRS-related skills. The interface supports search and filtering over runnable skills while exposing their domain assignments, verification status, and provenance information.
More in this topic

Agent Skill Learning

Agent skill learning develops reusable procedures and behaviors that help AI agents solve new tasks and adapt from experience.

See all 166 papers

Explore similar work

CardsList
Aug 30, 2026

Scientific Agent Skills: A Library of Procedural Knowledge for Research Agents

A language-model agent asked to analyse an experiment will usually return working code. Whether the analysis is defensible is a different question. A defensible analysis depends on procedural choices: which test the field accepts, which identifier namespace is authoritative, and which caveats must accompany a result. We present Scientific Agent Skills, an open library of 163 such procedures in 16 areas of practice, including genomics, cheminformatics, medical imaging, study design and scientific communication. Each skill is a directory built around a versioned, human-readable instruction file. An agent loads the file only when a task calls for it; the directory often also contains reference material and runnable scripts. We report no task-level evaluation and no host selection rate. We measure two properties of the documentation corpus: the always-resident descriptions of all 163 skills cost 7.1% of a 200,000-token window, and the median documented workflow fits within 23.9% of it, although 29 of 46 would overflow if every reference file were loaded. Openly licensed and available at https://github.com/K-Dense-AI/scientific-agent-skills.

Timothy Kassis, Vinayak Agarwal, Yuhuan He +2

K-Dense, Inc.

AI for ScienceLLM Agent Skill RetrievalScientific Workflow Automation
AskPDF
Mar 3, 2026

EvoSkill: Automated Skill Discovery for Multi-Agent Systems

Coding agents are increasingly used as general-purpose problem solvers, but their flexibility does not by itself confer the domain expertise needed for specialized tasks. Recent work addresses this through agent skills: reusable workflows, and code, that augment agents with domain-specific capabilities. Most skills today are hand-crafted, and existing evolutionary approaches optimize low-level artifacts (e.g. prompts & code) that are tightly coupled to specific models and tasks. We introduce EvoSkill, a self-evolving framework that automatically discovers and refines agent skills through iterative failure analysis. EvoSkill analyzes execution failures, proposes new skills or edits to existing ones, and materializes them into structured, reusable skill folders. A Pareto frontier of agent programs governs selection, retaining only skills that improve held-out validation performance while the underlying model remains frozen. We evaluate EvoSkill on two benchmarks: OfficeQA, a grounded reasoning benchmark over U.S.\ Treasury data, where it improves exact-match accuracy by 7.3% (60.6% →\to 67.9%); and SealQA, a search-augmented QA benchmark with noisy retrieval, where it yields a 12.1% gain (26.6% →\to 38.7%). We also investigate the zero-shot transfer capabilties of skills evolved on one task to the other; in particular: skills evolved from SealQA transfers zero-shot to BrowseComp, improving accuracy by 5.3% without modification demonstrating that skill-level optimization produces transferable capabilities beyond the training task.

Salaheddin Alzubi, Noah Provenzano, Jaydon Bingham +2

Sentient · Virginia Tech

Agent Skill LearningLLM Agent Self-ImprovementSelf-Evolving Agents
AskPDF
Jul 8, 2026

SkillCenter: A Large-Scale Source-Grounded Skill Library for Autonomous AI Agents

Autonomous AI agents can execute complex tasks with limited human review, yet they often lack the grounded operational knowledge to make their outputs not just executable but correct, secure, and maintainable. We introduce SkillCenter, to our knowledge the largest open skill library for agents by total count: 216,938 structured skills across 24 domain bundles. A SkillGate-filtered pipeline contributes 114,565 source-grounded skills from peer-reviewed journals, ArXiv, and over 24,000 technical sources, integrated with 102,373 community skills from GitHub and the ClawHub marketplace. We present the end-to-end framework that builds the pipeline subset: multi-source acquisition, an LLM-based quality gate (SkillGate), template-driven generation, iterative source-grounding, and quality-controlled publishing. Source grounding is a traceability guarantee: each retained claim maps to an exact quotation in its source. All skills ship as offline-searchable SQLite FTS5 bundles.

Tianming Sha, Yue Zhao, Lichao Sun +1

1Stony Brook University · University of Southern California · 3Lehigh University +1

LLM GroundingLLM Agent Skill LearningLLM Agent Skill RetrievalLLM Agents
AskPDF