Bias Assessor

Add bias/risk-of-bias assessment fields to an extraction table and populate them consistently. **Trigger**: bias, risk-of-bias, RoB, evidence quality, 偏倚评估,...

Security Scan

The skill's inputs, instructions, and required runtime (python) align with its stated purpose of adding risk-of-bias fields to an extraction table; nothing requested is disproportionate or unexpectedly powerful.

Purpose & Capability

The skill's name/description (add RoB fields to papers/extraction_table.csv) matches the runtime instructions and the declared requirement of python. The bundle also includes a large set of pipeline manifests and helper modules (tooling/*.py and many pipelines), which increases the attack surface compared with a minimal single-file helper but is coherent in the context of a research-pipeline collection.

Instruction Scope

SKILL.md explicitly limits inputs to papers/extraction_table.csv and outputs to an updated extraction_table.csv, prescribes a simple 3-level scale and short notes, and states 'Network: none'. The instructions do not ask the agent to read unrelated files, environment variables, or to transmit data externally.

Install Mechanism

No install spec is provided (instruction+bundled python modules). That is low-risk: nothing is downloaded at install time. The declared runtime requirement (python/python3) is appropriate for the included Python code.

Credentials

The skill requests no environment variables, credentials, or config paths. This is proportionate to a local CSV-editing utility.

Persistence & Privilege

The skill is not marked always:true and does not request elevated persistence or to modify other skills. The default ability for the agent to invoke the skill autonomously is unchanged and not by itself suspicious here.

Assessment

This skill appears to do what it claims: it reads papers/extraction_table.csv and populates conservative RoB fields (low|unclear|high) with short notes. Things to consider before installing: 1) Back up your papers/extraction_table.csv before running the skill (the tooling contains atomic write/backup helpers but always keep a copy). 2) The package includes many pipeline manifests and helper modules (larger code bundle than a tiny single-script helper); if you want minimal surface area, review or run only the specific script that edits extraction_table.csv. 3) Confirm you will run this in a workspace that does not contain sensitive files you don't want touched (the skill writes local files but declares no network access or credentials). 4) If you need stronger assurance, inspect tooling/quality_gate.py and the scripts that will be executed (scripts/run.py is referenced by the executor) to verify they only perform local CSV/file updates. 5) Human oversight is appropriate: review the produced rob_* fields and notes for errors or systematic drift.

Bias Assessor (risk-of-bias, lightweight)

Goal: make evidence quality explicit in a way that is quick, consistent, and auditable.

Inputs

• papers/extraction_table.csv

Outputs

• Updated papers/extraction_table.csv

Recommended fields

Use a simple 3-level scale (all lowercase): low | unclear | high.

Suggested columns to add (if missing):

• rob_selection
• rob_measurement
• rob_confounding
• rob_reporting
• rob_overall
• rob_notes

Workflow

1. Read papers/extraction_table.csv and identify the set of included studies.
2. If RoB columns are missing, add them (keep names stable once introduced).
3. For each study, fill each RoB domain:
   • low: design/reporting plausibly controls the bias
   • unclear: not enough information to judge
   • high: clear risk (e.g., missing controls, ambiguous measurement, selective reporting)
4. Set rob_overall conservatively:
   • high if any domain is high
   • unclear if no high but at least one unclear
   • low only if all domains are low
5. Add 1–3 short notes in rob_notes that justify the rating.

Definition of Done

• Every included paper row has all RoB columns filled.
• Values are strictly from low|unclear|high (no free-form scale drift).
• Notes are short and specific (what was missing / what was strong).

Troubleshooting

Issue: the table has mixed or inconsistent RoB column names

Fix:

• Normalize to the recommended column names and keep a single set across all rows.

Issue: the paper lacks enough methodological detail

Fix:

• Prefer unclear with a concrete note (“no details on X”) rather than guessing.