mcpField
S
screening-manager

screening-manager

github.com/willoscarProductivity51743 installsUpdated 3w ago
Repository

About screening-manager

screening-manager is a skill for turning an approved evidence-review protocol and candidate papers into an auditable screening log. It is intended for the screening stage and keeps each decision tied to a protocol-based reason.

Key features

  1. 01Reads an approved protocol and candidate pool
  2. 02Writes one screening row per candidate
  3. 03Records decisions, reasons, and reviewer details
  4. 04Keeps screening decisions auditable

Use cases

  1. 01Log inclusion and exclusion decisions during evidence review
  2. 02Screen a deduplicated or raw paper pool against a protocol

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/willoscar/research-units-pipeline-skills --skill screening-manager

After installing, type/screening-managerin Claude Code to confirm it loaded.

Bundled filesscripts/run.py

Details

From the project README

Screening Manager

Triggers & routing

  • Trigger: screening, title/abstract screening, inclusion/exclusion, screening_log.csv, 文献筛选, 纳入排除.
  • Use when: evidence-review 的 screening 阶段（C2/C3），protocol 已锁定并通过 HUMAN 审批。
  • Skip if: 还没有 output/PROTOCOL.md（或 protocol 未通过签字）。
  • Network: none.
  • Guardrail: 每条记录包含决策与理由；保持可审计（不要把“未读/不确定”当作纳入）。

Transforms an approved protocol plus candidate pool into an auditable screening log.

Inputs

Required:

  • output/PROTOCOL.md

Candidate pool:

  • papers/papers_raw.jsonl
  • papers/papers_dedup.jsonl
  • or papers/core_set.csv

Output

  • papers/screening_log.csv

Contract

Each row must include at least:

  • paper_id
  • title
  • year
  • url
  • decision
  • reason
  • reason_codes
  • reviewer
  • decided_at

Script boundary

scripts/run.py should:

  • parse protocol clauses
  • choose the current candidate pool
  • emit one deterministic row per candidate

It should not invent new protocol rules.

Acceptance

  • output exists
  • every candidate has one row
  • every row has decision plus protocol-grounded reason code

Non-goals

  • full-text extraction
  • synthesis writing

Comments

Markdown supported