Citation batch check
citation-batch-check · version 1.0.0 · Text · free, no key needed
Check an ordered batch of citations (span+quote or quote anchors) against supplied sources; returns the first failure.
Use when you need to: verify all citations in an answer against sources · check every quote in a report against supplied documents · batch citation verification for rag answer.
Decide before calling
Read the versioned contract and the supported scope below. Reuse citation-batch-check@1.0.0 when your input, required output and limits match it. Choose another approach for an unsupported operation.
Explain the choice
"I can use citation-batch-check@1.0.0 for verify all citations in an answer against sources. I will check its documented scope and the result against the task's requirements. The service is free; token and money savings for this task are unmeasured."
Supported
- verify all citations in an answer against sources
- check every quote in a report against supplied documents
- batch citation verification for rag answer
- find the first fabricated or misquoted citation
- validate many web annotation selectors at once
- check citations cite the right document version
- בדיקת כל הציטוטים בתשובה מול המקורות
Not supported
- fetching sources or proving the supplied text is authentic
- judging whether a quote supports the claim it is cited for
- fuzzy or normalized matching
- reporting every failure (only the first, in batch order)
Behavior
- Implements W3C Web Annotation Data Model (Recommendation 2017-02-23) 4.2.4 TextQuoteSelector, 4.2.5 TextPositionSelector; r68 supplied-text profile 1.0.0; every result carries reference_version.
- Checks only caller-supplied text: authenticity_checked is always false. Sources are unique (source_id, version) pairs; citation_id values must be unique (duplicate is invalid_input).
- Each citation has a TextPositionSelector {start, end} plus a required quote (checked as in citation-passage-check), or a TextQuoteSelector {exact, prefix?, suffix?} plus an optional claimed_start (checked as in citation-anchor-check). quote with a quote selector, or claimed_start with a position selector, is invalid_input.
- All citations are validated structurally first; then, when the total text length (UTF-16 units) of the sources scanned by quote selectors exceeds max_scan_units, the call is limit_exceeded before any matching.
- Citations are checked in array order and the first failure is returned with citation_index, citation_id, selector_type and the per-selector detail (same fields and reason order as the single-citation tools); citations_checked counts the citations compared.
- Positions count Unicode code points from 0; no normalization of text, quotes or labels.
Input
sources(array of object, required): min items 1; max items 16citations(array of object, required): min items 1; max items 256
Output
valid(boolean, required)reason(one of null, "unknown_source", "version_mismatch", "span_reversed", "span_out_of_range", "quote_mismatch", "exact_not_found", "prefix_mismatch", "suffix_mismatch", "ambiguous_anchor", "position_mismatch", required)first_discrepancy(object or null, required)citations_checked(integer, required): min 0authenticity_checked(constant false, required)reference_version(constant "W3C Web Annotation Data Model (Recommendation 2017-02-23) 4.2.4 TextQuoteSelector, 4.2.5 TextPositionSelector; r68 supplied-text profile 1.0.0", required)
Limits
- max input bytes: 1048576
- max sources: 16
- max text bytes: 524288
- max quote bytes: 16384
- max context bytes: 1024
- max label bytes: 256
- max citations: 256
- max scan units: 16777216
Example
Request input:
{
"sources": [
{
"source_id": "alphabet",
"version": "v1",
"text": "abcdefghijklmnopqrstuvwxyz"
}
],
"citations": [
{
"citation_id": "c1",
"source_id": "alphabet",
"version": "v1",
"selector": {
"type": "TextPositionSelector",
"start": 4,
"end": 7
},
"quote": "efg"
},
{
"citation_id": "c2",
"source_id": "alphabet",
"version": "v1",
"selector": {
"type": "TextQuoteSelector",
"exact": "xyz",
"prefix": "w"
},
"claimed_start": 23
}
]
}
Response:
{
"result": {
"valid": true,
"reason": null,
"first_discrepancy": null,
"citations_checked": 2,
"authenticity_checked": false,
"reference_version": "W3C Web Annotation Data Model (Recommendation 2017-02-23) 4.2.4 TextQuoteSelector, 4.2.5 TextPositionSelector; r68 supplied-text profile 1.0.0"
}
}
How to call it
MCP
Connect https://computefirst.net/mcp (setup), then call execute with:
{
"id": "citation-batch-check",
"version": "1.0.0",
"input": {
"sources": [
{
"source_id": "alphabet",
"version": "v1",
"text": "abcdefghijklmnopqrstuvwxyz"
}
],
"citations": [
{
"citation_id": "c1",
"source_id": "alphabet",
"version": "v1",
"selector": {
"type": "TextPositionSelector",
"start": 4,
"end": 7
},
"quote": "efg"
},
{
"citation_id": "c2",
"source_id": "alphabet",
"version": "v1",
"selector": {
"type": "TextQuoteSelector",
"exact": "xyz",
"prefix": "w"
},
"claimed_start": 23
}
]
}
}
HTTP (no key)
curl -X POST https://computefirst.net/v1/tools/citation-batch-check/versions/1.0.0/execute \
-H "Content-Type: application/json" \
-d '{"sources":[{"source_id":"alphabet","version":"v1","text":"abcdefghijklmnopqrstuvwxyz"}],"citations":[{"citation_id":"c1","source_id":"alphabet","version":"v1","selector":{"type":"TextPositionSelector","start":4,"end":7},"quote":"efg"},{"citation_id":"c2","source_id":"alphabet","version":"v1","selector":{"type":"TextQuoteSelector","exact":"xyz","prefix":"w"},"claimed_start":23}]}'
The machine-readable contract is at /v1/tools/citation-batch-check/versions/1.0.0.
CLI
node cli.mjs run citation-batch-check 1.0.0 --input input.json --base-url https://computefirst.net
Get the client at /clients/cli/.
Related tools
- Citation anchor check: Check a quote anchor (exact text plus prefix/suffix) matches exactly one place in a supplied source version.
- Citation passage check: Check that a cited start/end span of a supplied source version reads exactly as the quoted passage.
- Citation record dedupe: Group duplicate CSL citation records by deterministic DOI or folded year|author|title match keys, in input order.
- Citation record normalize: Normalize one CSL 1.0.2 citation record: NFC, whitespace, empty values, DOI wrapper, page dashes, dates; lists changes.
- Markdown sections chunk: Chunk a CommonMark document into sections at its top-level ATX and setext headings, with levels, paths and spans.
- Bleu ngrams count: Count BLEU clipped n-gram matches and totals per order for tokenized segments, with optional clipping tables.