K-Dense-AI/scientific-agent-skills

Scientific Agent Skills is a repository of portable, open-standard plugins designed for AI agents to perform scientific tasks. It provides a curated collection of specialized skills, including tools for protein experiment design and time-series machine learning, alongside a formalized development workflow with strict validation and security scanning. The system also includes automated utilities for generating technical diagrams and enforces comprehensive test coverage to ensure code quality across all contributions.

Added two new skills: Adaptyv Bio (v1.3) for protein experiment design and results retrieval via the Foundry API and Python SDK, and Aeon (v1.1) for scikit-learn compatible time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search.

skills · high confidence

A new Python script, \scripts/generate\_skill\_image.py\, has been added to the repository. This tool automates the creation of technical diagrams for skills by using the OpenRouter API. It operates in two stages: first, it uses a text model (Claude Opus) to analyze a skill's documentation and distill a description of its workflow; second, it sends that description to an image model (GPT-4o Image) to render a polished, flat-vector diagram. The script supports various aspect ratios and qualities, reads from a specified skill directory, and outputs the generated images to the \docs/images\ folder. It requires an \OPENROUTER\_API\_KEY\ environment variable or a \.env\ file.

scripts · high confidence

The project is renamed from 'claude-scientific-skills' to 'Scientific Agent Skills' and is now structured as a portable Agent Plugins 1.0.0 package (via \plugin.json\) that works with any AI agent supporting the open Agent Skills standard, not just Claude. This change introduces a formalized skill development workflow with strict \SKILL.md\ frontmatter validation (enforced by \strictyaml\), dedicated security scanning tooling (\scan\_skills.py\, \scan\_pr\_skills.py\), and comprehensive contributor guidance in \AGENTS.md\ and \CONTRIBUTING.md\. The repository also adds a Code of Conduct, a CITATION file, and an MIT license, while updating the version to 2.69.0.

(repo-wide) · high confidence

Added a shared test contract (\tests/\_contract\) that provides reusable test suites for CLI help output, structural validation, shared office/schematic code, and demo blocks, ensuring consistent quality across all skills. Additionally, a new meta-test suite (\tests/\_meta/test\_repo\_contract.py\) enforces that every skill shipping scripts has a corresponding test suite, preventing untested code from being merged.

tests · high confidence

This change introduces the core project configuration via pyproject.toml, establishing the 'scientific-agent-skills' package at version 2.69.0 with Python 3.13+ support and dependencies on cisco-ai-skill-scanner, pytest, and python-dotenv. It also adds specific requirement files for the Opentrons integration skill, pinning the library to version 9.1.1 for the 'flex' configuration and 9.0.0 for the 'ot2' configuration.

(dependencies) · high confidence

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

• First survey — no prior run to compare against. CAI 62.

• Code Health 74
• Architecture 100
• Maturity 73
• Readiness 46
• Security 83

• 300 commits — 273 feature/other, 27 fixes

• (root) — 84 commits
• (repo) — 25 commits
• docs/security-report.json — 10 commits
• docs/images — 8 commits
• skills/citation-management — 5 commits
• skills/database-lookup — 5 commits
• skills/genomic-intelligence — 4 commits
• skills/pi-agent — 4 commits
• docs/examples.md — 3 commits
• scientific-skills/simbad-database — 3 commits
• skills/imaging-data-commons — 3 commits
• skills/onekgpd — 3 commits
• skills/scanpy — 3 commits
• scientific-skills/exa-search — 2 commits
• scientific-skills/infographics — 2 commits
• scientific-skills/pacsomatic — 2 commits
• scientific-skills/pyhealth — 2 commits
• scientific-skills/research-grants — 2 commits
• scientific-skills/research-lookup — 2 commits
• skills/arbor — 2 commits

• fix: Close two gaps in the skill link contract, and fix what they surface (#258)
• fix: Fix Rowan examples for SDK 3.1.13 (#237)
• fix: Fix two install commands that point at packages not on PyPI (#261)
• fix: Merge pull request #116 from robotlearning123/fix/neuropixels-broken-links
• fix: Merge pull request #145 from xiaolai/fix/nlpm-uv-uv-pip-install
• fix: Merge pull request #146 from xiaolai/fix/nlpm-latchbio-uv-install
• fix: Merge pull request #164 from mvanhorn/fix/159-research-lookup-yaml-frontmatter
• fix: Merge pull request #210 from not-stbenjam/fix/markdown-mermaid-writing-broken-links
• fix: Merge pull request #213 from K-Dense-AI/fix/rdkit-api-corrections
• fix: Merge pull request #94 from jiaodu1307/fix/rdkit-skill-gzip-import-clean
• fix: Revert "chore: post-merge follow-ups for #252"
• fix: citation-management: fix BibTeX corruption and metadata defects
• fix: fix(database-lookup): retire dead PatentsView search API for ODP (#245)
• fix: fix(markdown-mermaid-writing): correct broken internal references in style guides
• fix: fix(rdkit): correct FractionCsp3 call and MolToSmarts parameter name
• fix: fix(rdkit): finish FractionCSP3 / CalcFractionCSP3 doc corrections (#244)
• fix: fix(simpy): report a non-UTF-8 resource CSV as unreadable, not as a bad integer (#259)
• fix: fix(stable-baselines3): repair broken API reference links (#233)
• fix: fix(usfiscaldata): correct nonexistent Treasury auctions_query field names (#201)
• fix: fix: address Skill Scanner findings - input sanitization for injection prevention
• …and 280 more

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

K-Dense-AI/scientific-agent-skills was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.