What makes biomedical-reference-verifier notable?

1.4K downloads. Trust evidence available

How should biomedical-reference-verifier be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

biomedical-reference-verifier

Verify or normalize biomedical and life-science reference lists, when the task is about AI-caused reference errors. Checks identifiers and bibliographic fields, preserves source evidence, and produces a searchable offline report. 生物医学/生命科学参考文献真实性验证skill，可以对参考文献列表（引文列表）进行多轮核查和错误修复，附带引文格式整理功能，能统一规范化所有引文为AMA、APA、GB/T 7714等格式。 Skill: biomedical-reference-verifier Owner: emp-tca Summary: Verify or normalize biomedical and life-science reference lists, when the task is about AI-caused reference errors. Checks identifiers and bibliographic fields, preserves source evidence, and produces a searchable offline report. 生物医学/生命科学参考文献真实性验证skill，可以对参考文献列表（引文列表）进行多轮核查和错误修复，附带引文格式整理功能，能统一规范化所有引文为AMA、APA、GB/T 7714等格式。 Tags: latest:1.2.1 Version history

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.4k

Updated

Oct 10, 2026

Version

1.2.1

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.4K downloads reported by the source. Last updated 10/10/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 10, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 10, 2026
Adoption signal: 1.4K downloadsadoption · observed Oct 10, 2026
Latest release: 1.2.1release · observed Sep 14, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s1762bkccamy86w1sawy1a2re58a8s0f:biomedical-reference-verifier

Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-emp-tca-biomedical-reference-verifier/snapshot"

Documentation

CLAWHUB

147,886 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: biomedical-reference-verifier
version: 1.2.1
description: "Verify or normalize biomedical and life-science reference lists, when the task is about AI-caused reference errors. Checks identifiers and bibliographic fields, preserves source evidence, and produces a searchable offline report. 生物医学/生命科学参考文献真实性验证skill，可以对参考文献列表（引文列表）进行多轮核查和错误修复，附带引文格式整理功能，能统一规范化所有引文为AMA、APA、GB/T 7714等格式。"
metadata:
openclaw:
requires:
bins:
- python3
envVars:
- name: NCBI_API_KEY
required: false
description: Optional NCBI API key for higher PubMed E-utilities rate limits.
- name: OPENALEX_API_KEY
required: false
description: Optional OpenAlex API key for authenticated metadata lookups.
---

# Biomedical Reference Verifier

Use this skill only for reference-list authenticity checking, AI-caused reference cleanup, and citation-format normalization. Do not use it to judge whether papers are relevant, suitable, or correctly used in the manuscript body unless the user explicitly asks for reference authenticity plus body cleanup.

## Core Rule

Always reduce the task to this sequence:

1. Build or read `biomedical-reference-verifier.records.v1`.
2. Choose the lightest pipeline.
3. Run the script.
4. Report result files and retained process files.
5. Ask before expensive manual recovery or process-file deletion.

Load `references/verification_policy.md` before classifying authenticity errors, changing severe items, or explaining network/query policy.

## Pipeline Choice

- **Format only**: use when the user only asks to convert or unify citation style. Run `--pipeline format-only`. Do not query Crossref/PubMed/OpenAlex.
- **Verify**: use when the user asks whether references are real, fake, wrong, AI-generated, DOI/PMID-mismatched, or metadata-corrupted. Run the default `verify` pipeline.
- **Verify then format**: use when the user asks both to check authenticity and standardize style. Run `verify` with the requested `--citation-style`; the fixed copy is generated after verification.

## Execution Mode

- **Fast** (`--mode fast`): use for quick scans and large routine lists. Stop after primary DOI evidence when available.
- **Balanced** (`--mode balanced`): default for ordinary authenticity checks. Auxiliary evidence lines receive a short 2.5-second grace period after primary DOI lookup completes (and supplied PMID lookup when enabled).
- **Strict** (`--mode strict`): use for final or pre-submission checks. Wait for every enabled evidence line to complete or fail explicitly.

Do not invent a local confidence score. For DOI-bearing references, resolve the DOI and directly compare returned title, authors, journal, year, DOI, and PMID with source fields. Keep the existing classification policy. Severe items are reported and retained; never auto-delete them.

Do not run AI-assisted per-paper web searching by default. Stop after batch verification and title recovery, then ask the user whether to continue.

README.md

# Biomedical Reference Verifier

Verify and normalize biomedical and life-science reference lists, with a focus on AI-caused citation errors.

The skill checks DOI, PMID, title, authors, journal, year, and other bibliographic metadata through Crossref, PubMed, and OpenAlex. It can also normalize references to AMA, APA, Vancouver, or GB/T 7714 style without overwriting the source document.

## Highlights

- Fast, Balanced, and Strict verification modes
- DOI-first batch verification and early DOI recovery
- Crossref, PubMed, and OpenAlex evidence
- Bounded retries and PubMed batch-failure isolation
- Detection of identifier hijacking and shifted identifiers
- Reusable audit indexes with lossless audit/index conversion
- Tolerant handling of user-edited prior artifacts
- Runtime metrics and request-event logging
- No hidden persistent cache
- Severe references are reported, never automatically deleted

## Requirements

- Python 3
- Standard library only
- Optional `NCBI_API_KEY` for higher PubMed limits
- Optional `OPENALEX_API_KEY`

## Quick start

```bash
python3 scripts/verify_references.py refs.md --mode balanced --output-dir reference-audit
```

Format without external verification:

```bash
python3 scripts/verify_references.py refs.md --pipeline format-only --citation-style ama
```

Run the offline regression suite:

```bash
python3 scripts/self_test_verify_references.py
```

See [SKILL.md](SKILL.md) for the complete workflow and [verification_policy.md](references/verification_policy.md) for classification and evidence rules.

## License

MIT-0

_meta.json

{
  "ownerId": "kn748sfft0aksxdrw9332r67w58a99n2",
  "slug": "biomedical-reference-verifier",
  "version": "1.2.1",
  "publishedAt": 1789392009429
}

references/verification_policy.md

# Verification Policy

## Evidence hierarchy

1. Crossref is the fastest and strongest default primary source for journal DOI, title, authors, journal, publisher, and year.
2. PubMed is the second evidence line for biomedical records, PMID mapping, and wrong PMID/title-pair detection.
3. OpenAlex is the third default evidence line for fast DOI/PMID external-ID corroboration and title recovery when Crossref is weak.
4. A DOI match from Crossref is the primary batch verification result. PubMed and OpenAlex matches can corroborate it, but should not block reporting.
5. Semantic Scholar, Europe PMC, bioRxiv/medRxiv, DataCite, and OpenCitations are backup channels, not default full-batch channels.
6. Search snippets, AI summaries, formatted APA/GB/T entries, and plausible journal names are not proof.
7. `biomedical-reference-verifier.records.v1` is the source-of-truth input layer. Source fields must come from the user's text; database-returned fields must stay in evidence/report/output objects.

## Error types

- `verified`: DOI/PMID or title search resolves to a canonical record and title, year, and first-author evidence agree.
- `verified_identifier_only`: DOI/PMID resolves to a canonical record, but the source title is missing or unreliable, so only the identifier is verified.
- `parser_error`: the input row is malformed or the parsed/source title is an author line, table header, metadata field, or other non-title text; stop before treating the row as a bibliographic conflict.
- `formatted_only`: formatting executed, authenticity not checked; not eligible for use on that basis.
- `minor_fix` / `minor_format_error`: the paper is real and metadata agree; only DOI casing, punctuation, URL style, initials, journal abbreviation, or minor year formatting needs correction.
- `partial_attribute_corruption`: the paper is real, but one or more attributes are corrupted, such as author, title, journal, year, volume/pages, DOI, or PMID.
- `identifier_hijacking`: DOI, PMID, or URL is real, but it points to a different paper than the reference text.
- `shifted_identifier`: DOI or PMID appears to belong to a nearby reference, commonly because one entry's DOI was attached to the previous or next citation.
- `semantic_hallucination`: the paper exists, but a surrounding claim or summary is not supported by that paper.
- `placeholder_generation`: the entry contains AI-like placeholder traces: generic title, invented DOI suffix, fake pages/volume, missing journal fields, or overly tidy but unverifiable metadata.
- `total_fabrication`: title, authors, journal, year, DOI/PMID, and searches fail to identify any canonical record.
- `unresolved`: automatic checks are insufficient; stop and ask before expensive manual recovery.

## Matching thresholds

- Title similarity >= 0.90: strong match.
- Title similarity 0.86-0.89: acceptable match if year and first author agree.
- Title similarity 0.78-0.85: possible match; mark as partial unless corroborated by DOI/PMID.
- Title si

CHANGELOG.md

# Changelog

## 1.2.1 — 2026-09-14

- Remove ambient email reads and omit contact email from requests unless explicitly supplied.
- Add selectable English HTML report template and localized generated eligibility messages.
- Document implementation locations for capability review; add privacy and language regression tests.

## 1.2.0 — 2026-09-14

Minor release following public version 1.1.3. SKILL.md and the CLI both identify this release as 1.2.0.

- Reject duplicate reference indices before verification or writing outputs, preventing silent loss or duplication of results.
- Reject output paths that alias the source, an explicitly reused artifact, or another output, including symlinks and hardlinks.
- Keep failed title queries distinguishable from completed searches with no match. Failed queries leave unresolved references unusable; successful empty searches retain the existing classification policy.
- Do not cache failed title searches as successful empty results. Detect malformed search responses and incomplete PubMed record retrieval.
- Include formatted-only entries in the Markdown detail report.
- Update the verification policy version to 2026-09-14.1 so older evidence is rechecked.
- Add 10 offline regression tests for these boundaries.
- Add the MIT-0 license and exclude Python caches and macOS metadata from publication.

Github ReposUpdated 20h agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.