verify-sources
Skill1 starsChanged 3 months agoMITView on GitHub
Verify that citations exist, resolve, and are what they claim to be. Use when the user wants to check citations, audit a reference list, confirm papers are real, or quality-check sources. Triggers on "verify sources", "check citations", "are these papers real", "audit references", "check my bibliography".
2 other projects use a version of this.
Adopted by
Most recently taken up by
- GoogleCloudPlatform/devrel-demostheir own version, first seen Nov 8, 2026
- bitwize-music-studio/claude-ai-music-skillstheir own version, first seen Jan 25, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- bitwize-music-studio/claude-ai-music-skills★ 668same name
- GoogleCloudPlatform/devrel-demos★ 404same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository13 other files it gives its agents.
- .claude/CLAUDE.mdCLAUDE.md
- factsSkill
- figureSkill
- full-reviewSkill
- ingestSkill
- onboardSkill
- plan-sectionSkill
- researchSkill
- reviewSkill
- save-materialSkill
- statusSkill
- teachSkill
- writeSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
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
- Distinctive: line 71
Never mark a citation verified because it looks plausible.
- Distinctive: line 73
Check at least two indexes before declaring a paper non-existent — coverage gaps are real, especially for books, theses, and non-English …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/verify-sources — Check every citation
Language models produce citations that look correct and do not exist. So do hurried authors working from memory. This skill checks each one against an index and reports what it could not confirm.
Input
A file path (a reference list, a research log, or a section) or a list of citations. If given a section, extract the in-text citations first.
For each citation
1. Existence
Try in this order and stop at the first confirmation:
- OpenAlex —
https://api.openalex.org/works?search={url_encoded_title}&per_page=5&mailto={email}, orhttps://api.openalex.org/works/doi:{doi}when a DOI is given. - Semantic Scholar —
https://api.semanticscholar.org/graph/v1/paper/search?query={title}&fields=title,authors,year,externalIds,venue,citationCount&limit=5(headerx-api-key). - Crossref —
https://api.crossref.org/works?query.bibliographic={citation}&rows=5&mailto={email}. - Web search as a last resort, against the publisher's own site.
VERIFIED requires the title, the first author, and the year to match. A one-year discrepancy is acceptable only for a preprint that was later published, and note it when it happens. A near-match on title with different authors is a different paper — the original is unverified.
2. Metadata
For verified sources, record the DOI, the actual venue, the publication year, the citation count, the type, and whether an open-access copy exists.
3. Tier
Tier 1 peer-reviewed journal or major conference; Tier 2 credible practitioner source or well-cited preprint; Tier 3 workshop or minor venue; Tier 4 unverifiable, working paper, or popular press.
4. Accuracy of use
Where the input is a section rather than a bare list, also check that the citation supports the sentence attached to it. Two failure modes matter and both are common: a real paper cited for a claim it does not make, and a real paper cited for a claim it makes about a different population or setting. Flag either, quoting the sentence.
Output
Save to docs/research/verification_<description>_<YYYY-MM-DD>.md:
# Source Verification: <description>
Date: <YYYY-MM-DD>
Input: <what was checked>
## Summary
Checked: X · Verified: X · Unverified: X · Tier 1: X · Tier 2: X · Tier 3: X · Tier 4: X
## Verified
| # | Citation | DOI | Venue | Cited by | Tier | Used in |
|---|----------|-----|-------|---------:|------|---------|
## Could not verify — action required
| # | Citation as given | What was searched | What was found instead |
|---|-------------------|-------------------|------------------------|
## Misattributed or overstated
| # | Citation | Sentence in the document | What the source actually says |
|---|----------|--------------------------|-------------------------------|
## Recommendation
<Which entries to remove, replace, or re-check by hand.>Rules
- Never mark a citation verified because it looks plausible. Search.
- Be most sceptical of very recent work, of sources with no DOI, and of anything the author does not remember reading.
- Check at least two indexes before declaring a paper non-existent — coverage gaps are real, especially for books, theses, and non-English work.
- Report failures plainly. An unverified citation left in a submitted document is an integrity problem, not a formatting one.
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
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.

