agentleFS
Sign inStart free
Google CloudKnown publisherGoogleCloudPlatform/LifeSciences/applications/open-agentic-science-engine/skills/citation-verification/SKILL.md

citation-verification

Skill53 starsChanged 55 days agoApache-2.0View on GitHub

Verify that citations in a document resolve to real records and that claimed titles match. Use before trusting a document's reference list.

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use when you need to verify whether citations in a document are real and accurate.

    line 10
  2. Distinctive:

    Do NOT use when:

    line 12

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

1. When to use, and when not

Use when you need to verify whether citations in a document are real and accurate.

Do NOT use when:

  • You need to resolve a single identifier to its record → use citation-resolution skill.
  • You need to search for literature by topic → use literature-search skill.

2. Preconditions

A file (JSON, Markdown, or text) containing citations or references.

3. Tool invocations

QuestionRunWrites to
Are the citations in this document real?oase cite verify <FILE>raw/literature/<slug>.citations.json
What is the verification assessment?oase cite analyze --from raw/literature/raw/literature/<slug>.analysis.json

4. Interpretation contract

Statuses

  • verified — citation resolved and title matched within tolerance.
  • suspect-title-match — citation resolved but title similarity is between suspect floor and match tolerance. Do NOT treat as verified.
  • phantom — citation did not resolve to any real record, or title completely mismatched.
  • unverified — could not check due to network error or timeout.

Verdicts

  • all-verified — every citation verified, none suspect or phantom.
  • suspect-matches-present — at least one suspect title match.
  • phantom-citations-present — at least one phantom citation.
  • partially-unresolved — some citations could not be checked.
  • no-citations-found — no extractable citations found.

Mandatory relays

Relay codeKindFires whenObligation
cite.phantom_citationDefectany phantomName the phantom references. A phantom citation invalidates the claim resting on it.
cite.suspect_title_matchQualifierany suspect-title-matchState that the reference resolved but the title did not match. Do not report it as verified.
cite.unresolved_offlineQualifierany unverified from network errorConfine verification claim to references that resolved.
cite.extraction_incompleteQualifierextraction_basis is not "structured"State that references were recovered by pattern match. Missed references were not checked.

Threshold set: citation-verification@1.0

ThresholdDefaultMeaning
title_match_tolerance0.75Boundary between verified and suspect
title_suspect_floor0.45Boundary between suspect and phantom
max_phantom_citations0Maximum acceptable phantom count
max_suspect_citationsUNRESOLVEDNot yet calibrated

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free