scienceworld-task-interpreter

This skill parses a user's high-level scientific task in the ScienceWorld environment and extracts the core objective and target location. It should be triggered when a new task instruction is received, especially those involving finding, comparing, or manipulating objects. The skill interprets the query to identify the goal (e.g., 'find the animal with the shortest life span') and any specified locations (e.g., 'animals are in the outside location'), outputting a clear, actionable sub-goal for navigation or observation.

Quality

82%

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

65%

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

A compact, well-structured instruction skill with concrete navigation/observation commands, but it is incomplete: the 'Output Format' section ends abruptly with no content, the execute step leans on abstract 'domain knowledge', and the existing action_glossary.md reference is never surfaced from the body.

Suggestions

Complete the truncated '**Output Format:**' section by specifying the exact Thought/plan structure Claude should emit before acting.

Replace the abstract 'Apply domain knowledge or comparative reasoning' with concrete execute-step guidance or worked examples for find/compare/manipulate tasks.

Link to references/action_glossary.md from the body (e.g., 'See [action_glossary.md](references/action_glossary.md) for the full action set') so the bundle file is clearly signaled rather than orphaned.

Dimension	Reasoning	Score
Conciseness	At 23 lines the body is lean, assumes Claude's competence, and avoids explaining what ScienceWorld is or how the environment works; every section carries instruction rather than filler.	3 / 3
Actionability	Gives concrete executable commands (`look around`, `teleport to LOC`, `focus on [OBJECT]`, `pick up [OBJECT]`), but the 'Analyze & Execute' step is abstract ('Apply domain knowledge or comparative reasoning') and the promised 'Output Format:' section is cut off with no actual format specified.	2 / 3
Workflow Clarity	The Parse -> Navigate -> Observe -> Analyze & Execute -> Output sequence is clearly ordered and includes a Thought-step confirmation checkpoint, but the final output step is truncated at the empty 'Output Format:' header, leaving the workflow's terminal artifact undefined.	2 / 3
Progressive Disclosure	The body is organized into clean sections and a bundle file references/action_glossary.md exists, but that glossary is never linked or signaled from the body, and the action commands it documents are instead inlined — references present but not clearly signaled.	2 / 3
	Total	9 / 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.

A well-constructed description: third-person voice, concrete actions, natural trigger terms, an explicit 'when' clause, and a clearly bounded niche. No fluff or over-claims, though slightly verbose.

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions — 'parses a user's high-level scientific task', 'extracts the core objective and target location', 'interprets the query to identify the goal', 'outputting a clear, actionable sub-goal' — each tied to the domain rather than vague abstractions.	3 / 3
Completeness	Explicitly answers both 'what' (parse/extract/interpret/output a sub-goal) and 'when' via the explicit trigger 'It should be triggered when a new task instruction is received, especially those involving finding, comparing, or manipulating objects'.	3 / 3
Trigger Term Quality	Covers natural task verbs a user would say — 'finding, comparing, or manipulating objects' — plus a realistic example phrasing ('find the animal with the shortest life span'), giving good coverage of natural trigger terms.	3 / 3
Distinctiveness Conflict Risk	Scoped to the specific 'ScienceWorld environment' with domain-locked triggers (scientific task, finding/comparing/manipulating objects), giving it a clear niche unlikely to fire for unrelated skills.	3 / 3
	Total	12 / 12 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: zjunlp/SkillNet
Path: experiments/src/skills/scienceworld/scienceworld-task-interpreter/SKILL.md
Commit: 7bb9a48

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.