Skills › Scientific Agent Skills › rdkit

rdkit

skill · 140 lines · BSD-3-Clause license · skills/rdkit/SKILL.md

The facts

Its description

Cheminformatics toolkit for fine-grained molecular control. SMILES/SDF parsing, descriptors (MW, LogP, TPSA), fingerprints, substructure search, 2D/3D generation, similarity, reactions. For standard workflows with simpler interface, use datamol (wrapper around RDKit). Use rdkit for advanced control, custom sanitization, specialized algorithms.

The description the agent reads to decide when to use it.

From its settingsline 3

Tools it may use, or may not

allowed-tools: Read Write Edit Bash

Tools it may use while it runs without asking you first.

From its settingsline 5

Files it mentions

references/core_capabilities.md, references/workflows_and_best_practices.md, references/review.md

Files the text names, such as ones to read or write. Found by matching words, so treat them as pointers.

Found by matching wordsline 42 line 60 line 73 line 123

Licence

BSD-3-Clause license, from the skill file

The terms its author shares it under, which let it be shown here.

From its settingsline 4 licence text

Other settings

compatibility: Tested with RDKit 2026.03.6 on Python 3.13; bundled scripts require the rdkit package. No credentials or network are needed after installation. Use conda-forge for the broadest binary support or PyPI package `rdkit` for supported platform wheels; `rdkit-pypi` is the legacy PyPI name. metadata: version: "1.5" last-reviewed: "2026-10-01" tested-rdkit: "2026.03.6" skill-author: K-Dense Inc.

More settings at the top of the file, shown as written.

From its settingsline 6 line 7

Other files in this skill (10) Close

The text

Reads a file Names a skill

Open on GitHub

Most marks come from matching words in the file, so treat them as pointers.

1---

2name: rdkit

3description: Cheminformatics toolkit for fine-grained molecular control. SMILES/SDF parsing, descriptors (MW, LogP, TPSA), fingerprints, substructure search, 2D/3D generation, similarity, reactions. For standard workflows with simpler interface, use datamol (wrapper around RDKit). Use rdkit for advanced control, custom sanitization, specialized algorithms.

4license: BSD-3-Clause license

5allowed-tools: Read Write Edit Bash

6compatibility: Tested with RDKit 2026.03.6 on Python 3.13; bundled scripts require the rdkit package. No credentials or network are needed after installation. Use conda-forge for the broadest binary support or PyPI package `rdkit` for supported platform wheels; `rdkit-pypi` is the legacy PyPI name.

7metadata:

8 version: "1.5"

9 last-reviewed: "2026-10-01"

10 tested-rdkit: "2026.03.6"

11 skill-author: K-Dense Inc.

12---

14# RDKit Cheminformatics Toolkit

RDKit Cheminformatics Toolkit

Raw lines 1 to 15

16## Overview

18RDKit is a comprehensive cheminformatics library providing Python APIs for molecular analysis and manipulation. This skill provides guidance for reading/writing molecular structures, calculating descriptors, fingerprinting, substructure searching, chemical reactions, 2D/3D coordinate generation, and molecular visualization. Use this skill for drug discovery, computational chemistry, and cheminformatics research tasks.

20**Tested stable baseline (reviewed 2026-10-01):** RDKit **2026.03.6**, distributed as `rdkit==2026.3.6` on PyPI; this is the current stable GitHub/PyPI release. Official installation docs continue to recommend conda-forge for most users, while cross-platform PyPI wheels are published under the `rdkit` package name. `rdkit-pypi` is the old PyPI package name and should only appear when maintaining legacy environments.

Overview

RDKit is a comprehensive cheminformatics library providing Python APIs for molecular analysis and manipulation. This skill provides guidance for reading/writing molecular structures, calculating descriptors, fingerprinting, substructure searching, chemical reactions, 2D/3D coordinate generation, and molecular visualization. Use this skill for drug discovery, computational chemistry, and cheminformatics research tasks.

Tested stable baseline (reviewed 2026-10-01): RDKit 2026.03.6, distributed as rdkit==2026.3.6 on PyPI; this is the current stable GitHub/PyPI release. Official installation docs continue to recommend conda-forge for most users, while cross-platform PyPI wheels are published under the rdkit package name. rdkit-pypi is the old PyPI package name and should only appear when maintaining legacy environments.

Raw lines 16 to 21

22## Installation and Setup

24Use `uv` when installing into an existing Python environment:

26```bash

27uv pip install "rdkit==2026.3.6"

28```

30For reproducible chemistry environments, especially when mixing compiled scientific packages, conda-forge remains the upstream recommendation:

32```bash

33conda create -c conda-forge -n my-rdkit-env rdkit

34conda activate my-rdkit-env

35```

37Avoid installing both conda `rdkit` and PyPI `rdkit`/`rdkit-pypi` into the same environment unless you are deliberately debugging packaging behavior. Mixed installs can make it unclear which binary extension is being imported.

Installation and Setup

Use uv when installing into an existing Python environment:

uv pip install "rdkit==2026.3.6"

For reproducible chemistry environments, especially when mixing compiled scientific packages, conda-forge remains the upstream recommendation:

conda create -c conda-forge -n my-rdkit-env rdkit
conda activate my-rdkit-env

Avoid installing both conda rdkit and PyPI rdkit/rdkit-pypi into the same environment unless you are deliberately debugging packaging behavior. Mixed installs can make it unclear which binary extension is being imported.

Raw lines 22 to 38

39## Core Capabilities

41Twelve capability areas, each with worked code, are documented in

42[references/core_capabilities.md](references/core_capabilities.md):

44| # | Area | Covers |

45| --- | --- | --- |

46| 1 | Molecular I/O and creation | SMILES, MOL files and blocks, InChI, SDF and SMILES suppliers, multithreaded reading, writers |

47| 2 | Sanitization and validation | disabling automatic sanitization, manual and partial sanitization, detecting problems first |

48| 3 | Analysis and properties | atom and bond iteration, ring information and SSSR, chirality and stereochemistry, fragments |

49| 4 | Descriptors | MW, LogP, TPSA, H-bond donors/acceptors, rotatable bonds, aromatic rings, bulk calculation, drug-likeness |

50| 5 | Fingerprints and similarity | topological, Morgan/ECFP via `rdFingerprintGenerator`, MACCS, atom pair, torsion, Avalon; Tanimoto and other metrics; Butina clustering |

51| 6 | Substructure searching | SMARTS queries, match retrieval, and a library of common patterns |

52| 7 | Chemical reactions | reaction SMARTS, applying reactions, reaction fingerprints |

53| 8 | 2D and 3D coordinates | depiction, template alignment, ETKDG embedding, force-field optimization, RMSD, constrained embedding |

54| 9 | Visualization | single and grid images, substructure highlighting, custom drawer options, Jupyter integration, fingerprint bit environments |

55| 10 | Molecular modification | explicit hydrogens, Kekulization, aromaticity, substructure replacement, charge neutralization |

56| 11 | Hashes and standardization | Murcko scaffold and canonical hashes, regioisomer hashes, randomized SMILES for augmentation |

57| 12 | Pharmacophore and 3D features | feature factories and feature extraction |

59Worked workflows and the performance, thread-safety, and version-sensitivity notes are in

60[references/workflows_and_best_practices.md](references/workflows_and_best_practices.md).

62Prefer portable exchange formats (SMILES, SDF) for shared data; for local caches RDKit's

63binary molecule representation avoids generic pickle.

Core Capabilities

Twelve capability areas, each with worked code, are documented in references/core_capabilities.md:

#	Area	Covers
1	Molecular I/O and creation	SMILES, MOL files and blocks, InChI, SDF and SMILES suppliers, multithreaded reading, writers
2	Sanitization and validation	disabling automatic sanitization, manual and partial sanitization, detecting problems first
3	Analysis and properties	atom and bond iteration, ring information and SSSR, chirality and stereochemistry, fragments
4	Descriptors	MW, LogP, TPSA, H-bond donors/acceptors, rotatable bonds, aromatic rings, bulk calculation, drug-likeness
5	Fingerprints and similarity	topological, Morgan/ECFP via `rdFingerprintGenerator`, MACCS, atom pair, torsion, Avalon; Tanimoto and other metrics; Butina clustering
6	Substructure searching	SMARTS queries, match retrieval, and a library of common patterns
7	Chemical reactions	reaction SMARTS, applying reactions, reaction fingerprints
8	2D and 3D coordinates	depiction, template alignment, ETKDG embedding, force-field optimization, RMSD, constrained embedding
9	Visualization	single and grid images, substructure highlighting, custom drawer options, Jupyter integration, fingerprint bit environments
10	Molecular modification	explicit hydrogens, Kekulization, aromaticity, substructure replacement, charge neutralization
11	Hashes and standardization	Murcko scaffold and canonical hashes, regioisomer hashes, randomized SMILES for augmentation
12	Pharmacophore and 3D features	feature factories and feature extraction

Worked workflows and the performance, thread-safety, and version-sensitivity notes are in references/workflows_and_best_practices.md.

Prefer portable exchange formats (SMILES, SDF) for shared data; for local caches RDKit's binary molecule representation avoids generic pickle.

Raw lines 39 to 64

Reads a file references/core_capabilities.md :42

Reads a file references/workflows_and_best_practices.md :60

65## Preserve chemical meaning

67Canonical isomeric SMILES records one molecular graph under a particular RDKit

68version. It does not merge tautomers, choose a pH-dependent protonation state,

69remove salts, or resolve unspecified stereo. Preserve the original structure and

70source ID before any explicit standardization policy. Similarity 1.0 is not an

71identity test: fingerprints collide and the bundled search ignores chirality

72unless `--chirality` is selected. See the tested identity/standardization example

73in [references/workflows_and_best_practices.md](references/workflows_and_best_practices.md).

75The helpers retain one-based source IDs through invalid-record filtering, reject

76empty molecules, and perform no implicit salt stripping or neutralization.

77SMILES files have no header; RDKit parses a SMILES/CXSMILES structure followed

78by an optional name. Text outputs retain enhanced stereo groups as CXSMILES;

79other CX annotations (such as drawing coordinates) are not an archive format.

80The search accepts one valid query molecule, not a query batch. MACCS always has

81167 positions (166 keys plus unused bit 0); `--bits` applies to other methods.

82The substructure helper's historical `pains` option is only five illustrative

83motifs, not the published PAINS catalogue. Invalid requested patterns abort.

84Descriptor cutoffs, QED, and alerts are research heuristics, not evidence of

85safety, potency, bioavailability, or synthetic feasibility.

Preserve chemical meaning

Canonical isomeric SMILES records one molecular graph under a particular RDKit version. It does not merge tautomers, choose a pH-dependent protonation state, remove salts, or resolve unspecified stereo. Preserve the original structure and source ID before any explicit standardization policy. Similarity 1.0 is not an identity test: fingerprints collide and the bundled search ignores chirality unless --chirality is selected. See the tested identity/standardization example in references/workflows_and_best_practices.md.

The helpers retain one-based source IDs through invalid-record filtering, reject empty molecules, and perform no implicit salt stripping or neutralization. SMILES files have no header; RDKit parses a SMILES/CXSMILES structure followed by an optional name. Text outputs retain enhanced stereo groups as CXSMILES; other CX annotations (such as drawing coordinates) are not an archive format. The search accepts one valid query molecule, not a query batch. MACCS always has 167 positions (166 keys plus unused bit 0); --bits applies to other methods. The substructure helper's historical pains option is only five illustrative motifs, not the published PAINS catalogue. Invalid requested patterns abort. Descriptor cutoffs, QED, and alerts are research heuristics, not evidence of safety, potency, bioavailability, or synthetic feasibility.

Raw lines 65 to 86

Reads a file references/workflows_and_best_practices.md :73

87## Common Pitfalls

891. **Forgetting to check for None:** Always validate molecules after parsing

902. **Sanitization failures:** Use `DetectChemistryProblems()` to debug

913. **Hydrogen representation:** Most 2D descriptors account for implicit H. Use `AddHs()` before embedding; it does not choose protonation at a specified pH.

924. **2D vs 3D:** Generate appropriate coordinates before visualization or 3D analysis. Check the conformer ID returned by `EmbedMolecule` before accessing coordinates: `-1` means embedding failed. For difficult molecules, enable `EmbedParameters.trackFailures` and inspect `GetFailureCounts()`; do not send a failed embedding into force-field optimization.

935. **SMARTS matching rules:** Remember that unspecified properties match anything

946. **Thread safety with MolSuppliers:** Don't share supplier objects across threads

Common Pitfalls

Forgetting to check for None: Always validate molecules after parsing
Sanitization failures: Use DetectChemistryProblems() to debug
Hydrogen representation: Most 2D descriptors account for implicit H. Use AddHs() before embedding; it does not choose protonation at a specified pH.
2D vs 3D: Generate appropriate coordinates before visualization or 3D analysis. Check the conformer ID returned by EmbedMolecule before accessing coordinates: -1 means embedding failed. For difficult molecules, enable EmbedParameters.trackFailures and inspect GetFailureCounts(); do not send a failed embedding into force-field optimization.
SMARTS matching rules: Remember that unspecified properties match anything
Thread safety with MolSuppliers: Don't share supplier objects across threads

Raw lines 87 to 95

96## Resources

Resources

Raw lines 96 to 97

98### references/

100This skill includes detailed API reference documentation:

101

102- `api_reference.md` - Comprehensive listing of RDKit modules, functions, and classes organized by functionality

103- `descriptors_reference.md` - Selected list of available molecular descriptors with descriptions

104- `smarts_patterns.md` - Common SMARTS patterns for functional groups and structural features

105

106Load these references when needing specific API details, parameter information, or pattern examples.

107

108Only the files listed in `references/` and `scripts/` are bundled local resources. Names such as `rdkit`, `datamol`, `scipy`, and `sklearn` refer to installable Python packages, not local files in this skill.

109

references/

This skill includes detailed API reference documentation:

api_reference.md - Comprehensive listing of RDKit modules, functions, and classes organized by functionality
descriptors_reference.md - Selected list of available molecular descriptors with descriptions
smarts_patterns.md - Common SMARTS patterns for functional groups and structural features

Load these references when needing specific API details, parameter information, or pattern examples.

Only the files listed in references/ and scripts/ are bundled local resources. Names such as rdkit, datamol, scipy, and sklearn refer to installable Python packages, not local files in this skill.

Raw lines 98 to 109

Names a skill datamol :108

110### scripts/

111

112Example scripts for common RDKit workflows:

113

114- `molecular_properties.py` - Calculate comprehensive molecular properties and descriptors

115- `similarity_search.py` - Perform fingerprint-based similarity screening

116- `substructure_filter.py` - Filter molecules by substructure patterns

117

118These scripts can be executed directly or used as templates for custom workflows.

119

scripts/

Example scripts for common RDKit workflows:

molecular_properties.py - Calculate comprehensive molecular properties and descriptors
similarity_search.py - Perform fingerprint-based similarity screening
substructure_filter.py - Filter molecules by substructure patterns

These scripts can be executed directly or used as templates for custom workflows.

Raw lines 110 to 119

120## Review evidence

121

122Release, official API/source links, runtime coverage, and explicitly illustrative

123examples are recorded in [references/review.md](references/review.md).

124

Review evidence

Release, official API/source links, runtime coverage, and explicitly illustrative examples are recorded in references/review.md.

Raw lines 120 to 124

Reads a file references/review.md :123

Boilerplate, lines 125 to 140. The same text appears in 125 other files here. Close

125## Citing Scientific Agent Skills

126

127This skill is part of Scientific Agent Skills by K-Dense. If it materially contributed to a

128manuscript, report, presentation, or code release, add the paper to the references or

129software section and tell the user you did so:

130

131> Kassis, T., Agarwal, V., He, Y., Patel, D., & Brueckner, A. M. (2026). Scientific Agent

132> Skills: A Library of Procedural Knowledge for Research Agents. arXiv:2609.00065.

133> https://doi.org/10.48550/arXiv.2609.00065

134

135Always cite the current version. The DOI and https://arxiv.org/abs/2609.00065 resolve to the

136latest arXiv version, so never append a version suffix such as `v1`. When network access is

137available, fetch https://arxiv.org/abs/2609.00065 (or

138http://export.arxiv.org/api/query?id_list=2609.00065) before writing the reference and take

139the author list, year, and version from that record. If the record lists a journal reference

140or publisher DOI, cite the published version instead.

From K-Dense-AI/scientific-agent-skills at 5f2bc03, 2026-10-26. Not read by a model yet. Summary and techniques by Claude Opus 5.5 on 2026-11-06.