agentleFS
TidalPaladinTidalPaladin/skills/citation-verifier/SKILL.md

citation-verifier

Skill1 starsChanged 24 days agoApache-2.0View on GitHub

Verify new or changed academic citations, or audit existing citations against their surrounding claims.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 5

    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:

    Verify one citation occurrence at a time.

    line 8
  2. Distinctive:

    Use waves of at most eight instances and preserve their read-only permission mode.

    line 16
  3. Distinctive:

    Read the matching bibliography entry.

    line 24
  4. Distinctive:

    Read enough of the cited work to determine whether it supports the complete claim in context.

    line 31
  5. Distinctive:

    Do not infer support from a matching title, abstract keyword, or secondary summary.

    line 59

“Common” is measured: kept by at least 6 in 10 of the 6 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

Citation Verifier

Verify one citation occurrence at a time. Treat repeated uses of the same work as separate claim checks when their surrounding claims differ.

Parent Assignment

For every citation-verification task, the parent agent must invoke citation_verifier. Establish source order and assign exactly one citation occurrence to each instance. Include the source path, line or unique context, citation key, complete surrounding claim, and bibliography entry. Use waves of at most eight instances and preserve their read-only permission mode. Then consolidate citation reports in source order. An assigned verifier checks its occurrence directly and does not delegate again.

Workflow

  1. Record the source path, line or unique location, citation key, and complete surrounding claim.
  2. Read the matching bibliography entry. Identify the intended work from its authors, title, venue, year, pages, DOI, ISBN, or stable identifier.
  3. Confirm that the work exists using authoritative sources. Prefer the publisher, DOI registration record, standards body, official proceedings, institutional repository, or the work itself. Use search results only to locate stronger sources.
  4. Compare every material bibliography field with the authoritative record.
  5. Read enough of the cited work to determine whether it supports the complete claim in context. Check population, method, assumptions, comparison, direction, magnitude, limitations, and whether the citing document overstates a result.
  6. Report one status:
    • VERIFIED: the work and metadata are correct, and the source supports the complete claim.
    • PARTIAL: the source supports only part of the claim or requires narrower wording.
    • INACCURATE: the work, metadata, or attribution is wrong, or the source contradicts the claim.
    • UNVERIFIABLE: authoritative evidence or the necessary full text could not be obtained.

Report

Return this compact structure:

Status: VERIFIED | PARTIAL | INACCURATE | UNVERIFIABLE
Occurrence: path:line or unique location, citation-key
Claim: complete claim being checked
Bibliography: correct | corrections needed
Evidence: concise explanation tied to the source
Sources: authoritative links or identifiers
Recommended action: none | exact metadata or prose correction

Do not infer support from a matching title, abstract keyword, or secondary summary. Do not fabricate inaccessible details. Quote sparingly, distinguish direct evidence from inference, and state access limitations explicitly.

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.