CtrlK
BlogDocsLog inGet started
Tessl Logo

boltz-small-molecule-adme

Predict Tier-1 ADME/ADMET for small molecules with Boltz from bare SMILES — no target, no docking. Use when the user wants solubility, permeability, or lipophilicity/logD for a molecule or list of molecules. Not for ranking molecules against a protein target (use boltz-small-molecule-screen, which already returns ADME free).

72

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Critical

Do not install without reviewing

SKILL.md
Quality
Evals
Security

Quality

Content

85%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.

A highly actionable, well-structured skill body with a validated workflow and clean reference split. The main weakness is redundancy — the sync/Codex behavior, price, 128-cap, and screen-skill redirect are each stated two or three times across the Workflow, Command Pattern, and Always Do This sections.

Suggestions

State the synchronous/no-background behavior (including the Codex session_id polling caveat) once — either in the Command Pattern comment or the Always Do This bullet — and drop the other two restatements.

Consolidate the 128-molecule cap and the $0.01/molecule pricing each into a single location (the Workflow step) and reference it from Always Do This instead of repeating the full details.

Merge the two boltz-small-molecule-screen redirects in the body (workflow intro and final bullet) into one, keeping the description's boundary statement as the other half of the pair.

DimensionReasoningScore

Conciseness

The body is dense and factual with no concept padding, but several facts are repeated: the synchronous/Codex-foreground behavior appears three times (step 4, the command comment, and the 'Always Do This' bullet), and the 128-molecule cap, the $0.01 price, cost confirmation, and the screen-skill redirect each appear twice. This fits the level 3 anchor ('mostly efficient but... could be tightened') better than level 4's 'minor instances'.

3 / 5

Actionability

The 'Command Pattern' block is copy-paste-ready with real flags (--model adme-v1, --idempotency-key, --root-dir), path and run-name conventions, the expected output path, and a concrete error code (adme_enumeration_failed). Covers estimate-cost, run, and result reporting, matching the fully-executable level 5 anchor.

5 / 5

Workflow Clarity

A numbered 5-step sequence with an explicit validation checkpoint (estimate-cost, show USD, wait for confirmation before submitting), per-molecule failure handling (status: failed, adme: null, report the error object), and recovery loops for missing CLI/auth (boltz-cli-setup then retry). The batch-operation cap-at-3 rule does not apply because validation and failure reporting are explicitly present.

5 / 5

Progressive Disclosure

The body stays an overview while the payload schema, output layout, and error handling are split into real, clearly signaled one-level-deep references (references/api.md, references/results.md), plus an Escape Hatch section with external links. Both referenced files exist in the bundle, matching the level 5 anchor.

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.

A strong description: concrete capabilities, an explicit 'Use when...' trigger clause, and explicit negative boundary guidance against the sibling screening skill. The only gap is a handful of natural trigger synonyms (drug-likeness, absorption, bioavailability).

DimensionReasoningScore

Specificity

Names multiple concrete capabilities — 'Predict Tier-1 ADME/ADMET', 'solubility, permeability, or lipophilicity/logD', 'from bare SMILES' — with comprehensive coverage of the skill's outputs in third-person voice. It exceeds the 'minor gaps' level 4 anchor because every endpoint and the input format are explicitly named.

5 / 5

Completeness

Explicitly answers both 'what' (predict ADME/ADMET endpoints for small molecules from SMILES) and 'when' via the concrete 'Use when the user wants solubility, permeability, or lipophilicity/logD' trigger clause. Matches the level 5 anchor's what-plus-when-with-triggers pattern.

5 / 5

Trigger Term Quality

'solubility', 'permeability', 'lipophilicity/logD', and 'molecule or list of molecules' are natural user phrases, but common variations such as 'drug-likeness', 'absorption', or 'bioavailability' are missing. Good keyword coverage with a few natural terms absent, matching the level 4 anchor.

4 / 5

Distinctiveness Conflict Risk

'Not for ranking molecules against a protein target (use boltz-small-molecule-screen...)' plus 'no target, no docking' explicitly disambiguate the nearest sibling skill and carve a clear niche with distinct triggers, matching the level 5 anchor.

5 / 5

Total

19

/

20

Passed

Validation

100%

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

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
openai/plugins
Reviewed

Table of Contents

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.