segment-anything-model

Foundation model for image segmentation with zero-shot transfer. Use when you need to segment any object in images using points, boxes, or masks as prompts, or automatically generate all object masks in an image.

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

Quality

Content

72%

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

The content is highly actionable with comprehensive executable code and good progressive disclosure to two real, shallow reference files. Its main weaknesses are mild verbosity from repeated examples and restated concepts, plus absent validation feedback loops in the batched/ONNX workflows.

Suggestions

Remove the duplicated single-point predict example in "Interactive segmentation" (it already appears in "Basic usage") and trim the "Key features" bullets and ViT architecture diagram that restate knowledge Claude already has.

Add an explicit validation/verification checkpoint to the Batched inference and ONNX deployment workflows (e.g., assert mask shape/non-empty before collecting, or verify the ONNX output matches the PyTorch reference) to introduce a feedback loop.

Consolidate the "Common issues" table with the dedicated references/troubleshooting.md reference to avoid restating the same issues inline and in the bundle.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean code-first content, but the basic-usage example is repeated in "Interactive segmentation > Point prompts", and "Key features"/"Common issues" prose plus the ViT architecture diagram restate concepts Claude largely already knows; it could be tightened, so it is not the fully lean score 3.	2 / 3
Actionability	Extensive copy-paste-ready executable code spans installation, checkpoints, SamPredictor, HuggingFace, all prompt types, automatic mask generation, and ONNX export/inference with real API calls and specific parameters, matching the fully-executable anchor.	3 / 3
Workflow Clarity	Sections are clearly sequenced by task, but the batched-inference and ONNX workflows lack explicit validation checkpoints or error-recovery feedback loops; per the rubric, missing validation in batch operations caps this at 2 rather than 3.	2 / 3
Progressive Disclosure	The body is an overview with two clearly-signaled, one-level-deep references (references/advanced-usage.md and references/troubleshooting.md), both verified present and containing no further nested .md links, matching the well-signaled one-level anchor.	3 / 3
	Total	10 / 12 Passed

Description

100%

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, third-person, and clearly answers both what the skill does and when to use it with concrete actions and natural trigger terms. It is distinct from other skills and free of vague fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "segment any object in images using points, boxes, or masks as prompts" and "automatically generate all object masks" — tied to specific prompt modalities, matching the multi-action anchor rather than the partial-action score 2.	3 / 3
Completeness	Explicitly states what ("Foundation model for image segmentation with zero-shot transfer") and includes an explicit "Use when you need to segment any object..." trigger clause, satisfying the both-what-and-when anchor; not score 2 because the trigger is explicit, not implied.	3 / 3
Trigger Term Quality	Uses natural terms a user would say ("image segmentation", "segment any object", "masks", "points, boxes") with good coverage; not merely technical jargon, so it sits at the natural-keyword anchor rather than score 2's "some relevant keywords".	3 / 3
Distinctiveness Conflict Risk	The SAM niche (zero-shot segmentation via points/boxes/masks) is distinctive and unlikely to trigger other skills, matching the clear-niche anchor rather than score 2's possible-overlap case.	3 / 3
	Total	12 / 12 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
skill_md_line_count	SKILL.md is long (501 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 2 missing	Warning
referenced_paths_exist	Referenced path issues: 1 missing	Warning

	Total	12 / 16 Passed

Repository: Orchestra-Research/AI-Research-SKILLs
Path: 18-multimodal/segment-anything/SKILL.md
Commit: 773a529

Reviewed: about 12 hours 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.