boltz

Structure prediction for protein, nucleic-acid, and small-molecule complexes with Boltz-2 (Passaro & Wohlwend et al. 2025, github.com/jwohlwend/boltz). Reach for this skill to validate designed binders against a target, to co-fold a protein with a SMILES or CCD ligand, or to get an open-source AlphaFold3 alternative with optional binding-affinity prediction.

Quality

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is an efficient, executable reference: complete input/output examples, confidence thresholds as pass lines, an error-symptom table, and guidance on accuracy-vs-memory tradeoffs. Minor tightening of positioning prose and an explicit step checklist would push it to the top of the scale.

Suggestions

Convert 'Running it' into a short numbered checklist (write YAML → run boltz predict → read confidence JSON → apply iptm/complex_plddt thresholds) to make the sequence and validation checkpoints explicit.

Trim the opening positioning paragraph ('closest in surface to AlphaFold3 ... pick chai1/openfold3/esmfold2 when ...') to one line, since the sibling-selection rationale adds tokens without action.

Move the version-sensitive detail ('Boltz v2.2.x caps affinity ligands at 128 atoms') into the Affinity head section explicitly flagged as version-bound, so a future version bump does not silently rot the guidance.

Dimension	Reasoning	Score
Conciseness	Largely lean and assumes Claude's knowledge (no 'what is a protein' filler), with every section earning its place; the positioning prose ('closest in surface to AlphaFold3 ... pick chai1 when ...') and the MSA paragraph could be trimmed slightly, so it sits just below the fully-lean anchor.	4 / 5
Actionability	Fully executable: a complete YAML input, a copy-paste `boltz predict complex.yaml --use_msa_server ...` command, concrete output paths and JSON fields, and an error table mapping symptoms to fixes — covering the common cases end to end.	5 / 5
Workflow Clarity	Clear sequence (write YAML → run predict → read confidence JSON → check thresholds) with explicit validation checkpoints (iptm > 0.5, complex_plddt > 0.7) and an OOM feedback path; it is prose- rather than checklist-formatted, leaving minor explicitness gaps versus the checklist anchor.	4 / 5
Progressive Disclosure	No bundle files exist and the skill is self-contained at ~95 lines with well-organized section headers and no nested external references, matching the simple-skill exception for scoring 5 on well-organized, reference-free content.	5 / 5
	Total	18 / 20 Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is concise, concrete, and tightly scoped: it states what Boltz-2 does, when to reach for it over siblings, and includes natural trigger phrases. It satisfies the what/when requirement cleanly with no fluff.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Structure prediction for protein, nucleic-acid, and small-molecule complexes', 'validate designed binders against a target', 'co-fold a protein with a SMILES or CCD ligand', and 'binding-affinity prediction' — giving comprehensive, non-abstract coverage.	5 / 5
Completeness	Explicitly answers both what ('Structure prediction ... with optional binding-affinity prediction') and when ('Reach for this skill to ...'), with concrete trigger phrases; voice is third person throughout.	5 / 5
Trigger Term Quality	Strong natural phrases a user would say ('validate designed binders against a target', 'co-fold a protein with a SMILES or CCD ligand', 'AlphaFold3 alternative') with good coverage; a few common synonyms (e.g. bare 'complex structure prediction') are absent, keeping it just below the comprehensive anchor.	4 / 5
Distinctiveness Conflict Risk	A clear niche (Boltz-2 co-folding) with distinct, specific triggers and minimal overlap risk; sibling models are differentiated in the body, anchoring a well-bounded trigger surface.	5 / 5
	Total	19 / 20 Passed

Validation

75%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 missing	Warning

	Total	12 / 16 Passed

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/claude-science/boltz/SKILL.md
Commit: 99df6e4

Reviewed: 4 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.