CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

rct-bias-assessment-rob2

Automates Risk of Bias 2 (ROB2) assessment for RCT papers by analyzing text against specific domains and synthesizing a report. Use when you need to assess the quality of a clinical trial paper or evaluate risk of bias.

Quality

71%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./scientific-skills/Evidence Insight/rct-bias-assessment-rob2/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

52%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.

The skill's core assessment workflow is well structured with a clear domain sequence, explicit synthesis rules, and a concrete JSON output contract, and its reference file and scripts genuinely exist. The main weaknesses are a large tail of generic, self-duplicating template sections that consume tokens without skill-specific value, and helper-script documentation that is inaccurate versus the actual scripts (missing required argument, wrong output filename).

Suggestions

Cut or merge the generic tail sections (When to Use, When Not to Use, Required Inputs, Output Contract, Failure Handling, User Checkpoints, Input Validation, Quick Validation) into a short, single section, keeping only ROB2-specific guidance; the missing-input rule is currently repeated in at least four places.

Correct the PDF extraction instructions to match the script: show `python scripts/extract_pdf.py paper.pdf` (the path argument is required) and fix the documented output to `<stem>_fulltext.txt` rather than `full_text.txt`; also use the `scripts/` prefix consistently for both helper scripts.

Embed validation as an explicit checkpoint in the workflow (e.g., after synthesizing the overall judgement, verify all five domain fields and Overall are set and consistent with the High/Some-concerns/Low rules before emitting the JSON) instead of leaving it only in a trailing checklist.

DimensionReasoningScore

Conciseness

The core sections (Usage, Workflow Steps, Output Format, Helper Scripts) are tight, but roughly 40% of the body is generic template boilerplate ("When to Use", "When Not to Use", "Required Inputs", "Output Contract", "Failure Handling", "User Checkpoints", "Input Validation", "Quick Validation") that is not specific to ROB2 and duplicates itself (e.g., the missing-input rule appears in at least four sections). This is noticeably verbose with several padded sections — anchor 2 — rather than mostly efficient with only some trimmable explanation (anchor 3).

2 / 5

Actionability

The JSON output contract and domain workflow steps are concrete, but the helper-script guidance is not executable as written: the body shows `python extract_pdf.py` with no arguments, while the actual script requires a PDF path argument and writes to `<stem>_fulltext.txt`, not the documented `full_text.txt`. That is concrete guidance with missing/incorrect key details — anchor 3 — rather than mostly executable guidance with only minor gaps (anchor 4).

3 / 5

Workflow Clarity

The sequence is clear: extract study info, assess the five named domains, synthesize via explicit High/Some-concerns/Low rules, output to a specified JSON schema, with a separate Quick Validation section and failure-handling guidance. It falls short of anchor 5 because validation is a trailing checklist rather than embedded checkpoints with an explicit validate → fix → re-validate feedback loop, and phrases like "assesses each of the 5 ROB2 domains in parallel (conceptually) or sequentially" introduce ambiguity; this is not a destructive or batch operation, so the workflow-clarity cap of 3 does not apply.

4 / 5

Progressive Disclosure

The body is well sectioned, and its references are real, one level deep, and clearly signaled ("Refer to [rob2_guidelines.md](references/rob2_guidelines.md)" and `scripts/assess_rob2.py` both exist in the bundle). Minor gaps keep it below anchor 5: `extract_pdf.py` is referenced without its `scripts/` path (inconsistent with `scripts/assess_rob2.py` styling), and its documented output filename contradicts the script's actual behavior, slightly confusing navigation to the right tool.

4 / 5

Total

13

/

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 that clearly and explicitly states both what the skill does and when to use it, with natural trigger phrases and a well-defined niche. The main gaps are modest: it could enumerate the five assessment domains and add common synonyms like "critical appraisal" to sharpen triggers and distinctiveness against sibling risk-of-bias tools.

DimensionReasoningScore

Specificity

The description names its domain clearly ("Risk of Bias 2 (ROB2) assessment for RCT papers") and lists 1-2 concrete actions ("analyzing text against specific domains and synthesizing a report"), but does not enumerate the domains assessed or the output form, so coverage is not comprehensive. It fits anchor 3 rather than 4 because the actions are fewer and less enumerated than the anchor-4 pattern of several distinct concrete actions.

3 / 5

Completeness

It explicitly answers both questions: the "what" ("Automates Risk of Bias 2 (ROB2) assessment for RCT papers by analyzing text against specific domains and synthesizing a report") and the "when" ("Use when you need to assess the quality of a clinical trial paper or evaluate risk of bias") with concrete trigger phrases. This matches the anchor-5 example structure of what + explicit use-when triggers; voice is third person so no voice penalty applies.

5 / 5

Trigger Term Quality

Natural trigger phrases users would actually say are present: "assess the quality of a clinical trial paper" and "evaluate risk of bias", plus domain terms RCT and ROB2. It misses common synonyms such as "critical appraisal" or "methodological quality", so keyword coverage is good but not comprehensive, matching anchor 4 rather than 5.

4 / 5

Distinctiveness Conflict Risk

The niche is clear (ROB2 for randomized controlled trials) with distinct triggers around clinical trial quality, making it mostly distinguishable from unrelated skills. However, the trigger phrase "risk of bias" is shared with closely related tools (e.g., ROBINS-I for non-randomized studies, other appraisal instruments), leaving minor overlap risk with sibling appraisal skills — anchor 4 rather than 5.

4 / 5

Total

16

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

aipoch/medical-research-skills

Path

scientific-skills/Evidence Insight/rct-bias-assessment-rob2/SKILL.md

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.