deepspot-m

Generate transcriptome-wide virtual spatial transcriptomics from H&E histology with DeepSpot-M. Use when you need spatial gene expression in log1p-CPM for 224x224 tiles at about 20x, want to query protein-coding genes by symbol instead of a fixed panel, or want to run prediction across a whole slide after tiling with histolab.

Quality

86%

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

The canonical home for this skill is deepspot-m in K-Dense-AI/scientific-agent-skills

Quality

Content

80%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 well-structured, highly actionable body with copy-paste code and clean progressive disclosure to two real reference files. The main gap is workflow clarity: the batched whole-slide workflow lists steps but does not weave an explicit validate/verify checkpoint into the sequence, so it sits at the rubric's batch-operation cap of 3.

Suggestions

Add an explicit validation/verification step to the whole-slide workflow (e.g. '4. Verify: check the tiles-by-genes matrix has no all-NaN rows and every tile has coordinates before assembling AnnData; re-run failed batches if found'), turning it into a validate→fix→retry loop.

Trim over-explanation Claude already knows, such as 'unsqueeze(0) adds the batch dimension' and the LoRA/cross-attention/router architecture paragraph in the Overview, unless those details change how the skill is invoked.

Fold the existing require_tile size check and unknown-symbol KeyError handling into the workflow steps as named checkpoints so validation is part of the sequence rather than a separate section.

Dimension	Reasoning	Score
Conciseness	Mostly lean with executable signal density, but includes a few explanations Claude already knows such as 'unsqueeze(0) adds the batch dimension' and the model-architecture paragraph in the Overview that goes beyond what is needed to use the skill.	4 / 5
Actionability	Fully executable, copy-paste-ready guidance throughout: install command, huggingface-cli login, from_pretrained/predict_genes calls, a require_tile validator, a load_deepspotm lazy-import helper, and an embedding-source table covering the common cases.	5 / 5
Workflow Clarity	The whole-slide workflow is a clear numbered sequence (tile with histolab, stack batches, predict_genes per batch, concatenate into a matrix), but the batch workflow itself has no explicit validation/verification checkpoint inside the steps; per the rubric, a batch operation missing validation caps this at 3 even though tile-size and unknown-symbol checks exist elsewhere.	3 / 5
Progressive Disclosure	Clear overview with well-signaled, one-level-deep references — 'references/api.md' and 'references/whole_slide.md' (both verified to exist) — and a 'Detailed references' section stating what each file contains; core usage is inline while the full API and slide-scale loop are split out.	5 / 5
	Total	17 / 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, third-person description that concretely states what the skill does and gives explicit, multi-clause 'Use when' trigger guidance tied to specific units, tile sizes, and tooling. Its only weak spot is trigger-term breadth, missing a few common synonyms/abbreviations a user might say.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Generate transcriptome-wide virtual spatial transcriptomics from H&E histology', 'query protein-coding genes by symbol instead of a fixed panel', and 'run prediction across a whole slide after tiling with histolab' — giving comprehensive coverage of what the skill does.	5 / 5
Completeness	Explicitly answers both what ('Generate transcriptome-wide virtual spatial transcriptomics from H&E histology with DeepSpot-M') and when ('Use when you need spatial gene expression in log1p-CPM... want to query protein-coding genes by symbol... or want to run prediction across a whole slide') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Good domain keyword coverage ('spatial gene expression', 'spatial transcriptomics', 'H&E histology', '224x224 tiles', '20x', 'whole slide', 'tiling with histolab') that a user in this niche would naturally say, but a few common synonyms/abbreviations (e.g. 'ST', 'Visium', 'spatial omics') are absent.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche — virtual spatial transcriptomics from H&E via DeepSpot-M with named tooling (histolab) and units (log1p-CPM) — making overlap with other skills minimal and triggers distinct.	5 / 5
	Total	19 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/biology/deepspot-m/SKILL.md
Commit: 4082a2e

Reviewed: 1 day 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.