Agent skill K-Dense Inc. 2025
Scientific Agent Skills: reusable research procedures, including literature review
A repository of reusable scientific task procedures for compatible AI agents. Review here covers the collection README and the literature-review SKILL.md, not an audit of every skill or an empirical test of research gains.
Workflow level n/a Source: Repository-documented Outcome: Not empirically evaluated here
Research applicability
The reviewed literature-review procedure covers planning, multi-database search, screening, extraction, synthesis, citation checks and document preparation. Other collection categories include scientific coding, analysis and communication as documented by maintainers.
Getting started
Read the selected SKILL.md and linked helpers as untrusted code/instructions; pin a reviewed revision and test in an isolated environment with non-sensitive data. For literature review, define databases and eligibility criteria before searching, and manually verify supporting sources.
Maintenance
GitHub API checked 2026-10-15: canonical repository is K-Dense-AI/scientific-agent-skills (the former claude-scientific-skills URL redirects); not archived, last pushed_at 2026-10-12. Snapshot only, not a code/security audit.
License
MIT license verified in LICENSE.md and in the reviewed literature-review skill metadata. Third-party packages, data and service access remain separately licensed.
Related resources
Workflow & human oversight
A library of instructions and optional helpers is not an autonomous workflow by itself. Level depends on the host agent, selected skill, granted tools and human checkpoints. Researchers must inspect skills before installation and approve data access, execution and final claims.
Workflow level is not an evidence score; higher is not better evidence.
Problem
Research agents need explicit task procedures, provenance requirements and validation boundaries rather than generic prompts alone.
Approach
Skills package task instructions and may include references, templates and executable helpers. The reviewed literature-review skill distinguishes ranked discovery from comprehensive coverage and DOI registration from checking whether a source supports a claim.
Limitations
Maintainer documentation, not an independently tested claim of greater accuracy or faster research. Skill instructions can cause code execution, network requests and file changes in their host. The entire collection, dependencies and scripts were not audited; no downloaded skill was installed or executed.
Reproducibility
The retrieved README, literature-review skill and LICENSE are preserved in discovery/review. Review the exact version and its dependencies before adoption; optional external services can change.
Adoption requirements
An Agent Skills-compatible host, narrowly scoped permissions, researcher oversight and task-specific dependencies. The reviewed literature-review skill declares Python 3.10+ with requests; optional PDF export requires Pandoc/XeLaTeX and optional discovery/generation services need their own access.
Comments
No comments yet.
Add a comment
Comments are held for moderation before appearing. Critical or corrective comments are welcome — especially evidence that an entry overstates what has been verified.