CtrlK
BlogDocsLog inGet started
Tessl Logo

scienceworld-task-parser

Analyzes user instructions in ScienceWorld environments to extract specific task requirements and constraints. Use when receiving a new task to identify required objects, target locations, and action sequences before taking any environment actions.

64

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./experiments/src/skills/scienceworld/scienceworld-task-parser/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

71%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 lean, well-sequenced instruction skill with concrete commands and a good worked example, undermined by broken navigation: the bundled references/task_patterns.md is orphaned (never linked from the body) and 'the trajectory' is a dangling reference. Linking the reference file and making the verification step explicit would lift the two weakest dimensions.

Suggestions

Add a clearly signaled, one-level-deep link to the bundled reference, e.g. under a 'Task patterns' section: 'See [references/task_patterns.md](references/task_patterns.md) for common instruction templates (find-and-move, activate, mix) and the object classification guide.'

Fix the dangling pointer in the 'Clarity' principle ('Structure your internal reasoning using the "Thought:" prefix... as shown in the trajectory') — either point to a real file or drop the 'as shown in the trajectory' clause.

Strengthen 'Verification' into an explicit checkpoint (e.g., after `move OBJ to CONTAINER`, run `look around` / `examine CONTAINER` and confirm the object is present before declaring the task complete).

DimensionReasoningScore

Conciseness

The ~44-line body is efficient and assumes competence — e.g., 'All containers are pre-opened. Do not use `open` or `close`' and tight command bullets — with no explanations of known concepts. Minor trimmable redundancy remains (teleport guidance appears in both section 2 and 'Directness', and the 'Verification' principle about repeating `look around` is near-trivial), placing it at anchor 4 rather than the fully lean anchor 5.

4 / 5

Actionability

Concrete environment commands are given throughout (`look around`, `teleport to LOC`, `examine OBJ`, `focus on OBJ`, `move OBJ to OBJ`) and the worked example ('teleport to workshop' → 'focus on battery' → 'move battery to purple box') is executable as written. However, step 2 of Execute ('Execute the primary action from the parsed instruction') stays at the template level and more varied concrete command patterns are left to the unlinked reference file, so it falls just short of anchor 5's copy-paste-ready coverage of common cases.

4 / 5

Workflow Clarity

The four-phase sequence (Parse → Survey → Identify → Execute with 'Signal Intent' before the core action) is clearly ordered, and the 'Verification' principle plus 'Use `examine OBJ`... if you need more detail' provide some checkpoints. Validation is soft ('a second `look around` is acceptable') rather than an explicit validate-and-recover loop, matching anchor 4's 'most checkpoints present; minor validation gaps' rather than anchor 5.

4 / 5

Progressive Disclosure

The body is well-sectioned and appropriately short, but the bundle contains references/task_patterns.md (a catalog of task patterns and an object classification guide) that is never referenced from SKILL.md, and the body's 'as shown in the trajectory' pointer resolves to no file present in the bundle. Scoring against the actual bundle structure, existing references that are not signaled from the skill body matches anchor 3 ('references present but not clearly signaled') rather than anchor 4.

3 / 5

Total

15

/

20

Passed

Description

78%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 third-person capability statement paired with an explicit 'Use when' trigger and a distinctive ScienceWorld niche. The 'when' clause could be slightly more specific with varied trigger phrasing, but it clearly communicates both what the skill does and when to invoke it.

DimensionReasoningScore

Specificity

The description lists several concrete actions — 'Analyzes user instructions... to extract specific task requirements and constraints' and 'identify required objects, target locations, and action sequences' — with only minor coverage gaps (e.g., it does not mention the action-planning output). It names multiple specific capabilities but is not fully comprehensive, matching anchor 4 rather than 5.

4 / 5

Completeness

Both parts are present: a clear third-person 'what' ('Analyzes user instructions... to extract specific task requirements and constraints') and an explicit 'Use when receiving a new task...' clause. The 'when' is explicit but somewhat generic ('receiving a new task') rather than offering concrete varied trigger phrases, matching anchor 4 ('when' could be more explicit or specific) rather than the fully concrete anchor 5.

4 / 5

Trigger Term Quality

Natural trigger phrases like 'receiving a new task', 'task requirements', 'objects', 'target locations', and 'action sequences' are present and would be said by a user in this context, alongside the distinctive 'ScienceWorld' keyword. A few natural variations (e.g., 'instruction', 'goal', 'mission') are missing, so it sits at anchor 4 rather than 5.

4 / 5

Distinctiveness Conflict Risk

'ScienceWorld environments' carves out a clear niche with distinct triggers; no other skill category would plausibly match 'receiving a new task to identify required objects, target locations, and action sequences' in that domain. This clearly matches anchor 5 (clear niche, minimal conflict risk), not anchor 4 which requires some overlap risk with closely related skills.

5 / 5

Total

17

/

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
zjunlp/SkillNet
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.