Update to Pendar

SKILL.md text changes

− removed · + added

--- before
+++ after
@@ -1,79 +1,100 @@
 ---
 name: verify-citations
-description: Check whether cited papers actually exist, are described correctly, and have not been retracted. Use when the user pastes a bibliography, reference list, or citations from a draft, asks whether a paper is real, or wants references from an AI-generated answer checked before relying on them.
+description: Check whether a cited paper actually exists, is described correctly, and has not been retracted. Use when the user asks whether a specific paper is real, pastes one or two references, or wants a citation from an AI-generated answer checked before relying on it. For a whole reference list, use audit-a-bibliography.
 metadata:
-  short-description: Confirm papers exist, match their citation, and are not retracted
+  short-description: Confirm a paper exists, matches its citation, and is not retracted
 ---
 
 # Verify citations
 
-Confirm each cited paper exists in the published record, that the citation
+Confirm a cited paper exists in the published record, that the citation
 describes it accurately, and that it has not been retracted.
 
 A citation that cannot be found is the finding. Report it as unverified and stop
-there — never substitute the closest real paper, and never fill in a missing
-DOI, year, or venue from memory. A confidently wrong reference is worse than an
-acknowledged gap, because the user will submit it.
+there — never substitute the closest real paper, and never supply a title, page,
+year, venue, or DOI from memory. A confidently wrong reference is worse than an
+acknowledged gap.
+
+Check the citation even when the paper is familiar. The user is asking what the
+record says, and a well-known paper can still be cited with the wrong year,
+venue, or authors, or have been retracted since.
 
 ## Quick start
 
-1. Split the input into individual citations. One `verify_paper` call per
-   citation, never several at once in one string.
-2. Pass whatever the user gave: a DOI, a bare title, or a full citation string
-   all work. A `paper_id` from an earlier Pendar result also resolves exactly.
-3. Classify each result — verified, ambiguous, retracted, or not found — using
-   the rules below.
-4. Report every citation in the order given, then a one-line tally.
+1. One `verify_paper` call per citation. For more than two or three, use
+   `audit_bibliography` instead (see `audit-a-bibliography`).
+2. Pass whatever the user gave, exactly as written: a DOI, an arXiv or PubMed ID,
+   a bare title, a full citation string in any common style, or a journal-style
+   abbreviated reference such as
+   `E. Brown and H. M. Jaeger, Phys. Rev. Lett. 103, 086001 (2009).` Do not tidy
+   it up first; a `paper_id` from an earlier Pendar result also resolves exactly.
+3. Classify the result — verified, ambiguous, retracted, mismatch, or not found —
+   using the rules below.
+4. If the call returns an error saying the paper index did not respond, the
+   citation was **not checked**. Try once more; if it fails again, say it could
+   not be checked right now. Never report it as not found.
 
 ## Reading the result
 
-- `found: false` — the paper could not be resolved. Say so plainly and quote
-  `reason`. Do not offer a different paper as though it were the one cited. If
-  the user wants alternatives, offer to search the literature instead, as a
-  separate step they asked for.
-- `candidates` non-empty — the citation was too ambiguous to resolve. List the
-  candidates with year, venue, and author so the user can choose. Ask which one
-  they meant. Do not pick the top candidate silently.
-- `match_confidence` below about 0.8 — the match is uncertain even though a
-  paper came back. Show the resolved `title`, `year`, and `authors` next to what
-  the user wrote so they can judge it themselves.
+- `found: true` — verified, unless `metadata_mismatch` is also true (below).
+  `resolved_by` says how:
+  - `doi`, `paper_id` or `pmid` — exact identifier match. An arXiv ID resolves
+    as `doi`.
+  - `title_search` — the title matched; check `match_confidence`.
+  - `biblio` — an abbreviated reference with no title, matched on its journal
+    locator. `matched_on_fields` lists what agreed (year, volume, page, first
+    author or journal). This is a full verification. The canonical title is the
+    one in `paper.title`; quote that, not a title recalled from memory.
+- `found: true` with `metadata_mismatch: true` — the paper exists, but the cited
+  title differs from the record: found at the journal, volume and page given, by
+  the cited first author. `reason` quotes the record's title. Report it as a
+  citation to correct, with the published title.
+- `match_confidence` below about 0.8 — uncertain even though a paper came back.
+  Show the resolved `title`, `year`, and `authors` next to what the user wrote.
+- `candidates` non-empty — too ambiguous to resolve. List the candidates with
+  year, venue, and author, and ask which one was meant.
+- `found: false` — quote `reason`. Do not offer a different paper as though it
+  were the one cited, and do not fill in missing details yourself.
 - `retraction_warning` non-empty, or `paper.is_retracted: true` — lead with the
-  retraction. Say the paper is retracted before saying anything about its
-  findings, and advise against citing it as standing evidence.
-- `paper.work_type` of `preprint` — flag that it has not been peer reviewed. The
-  user may have cited it as a journal article.
-- Metadata that disagrees with the user's citation — a different year, venue, or
-  author list — report the discrepancy and give the canonical values from
-  `paper`. Do not quietly correct it; the user needs to know their source was
-  wrong.
+  retraction and advise against citing it as standing evidence.
+- `paper.work_type` of `preprint` — flag that it has not been peer reviewed.
+- `is_paratext: true` — the record is an erratum, correction notice, or similar,
+  not the research article itself. Say so.
+- `venue` — facts about the journal: whether it is open access, listed in the
+  open-access journal directory, and has an ISSN and a publisher. Report these as
+  facts. Do not call a journal predatory or low quality.
+- `paper.impact_percentile` — where the paper's citation count sits among papers
+  published the same year. Say "among papers from the same year".
+- Metadata that disagrees with the citation — a different year, venue, or author
+  list — report the discrepancy and give the canonical values from `paper`.
 
 ## Never
 
-- Invent or guess a DOI, a `paper_id`, or a URL. Use `paper.doi` and
-  `paper.landing_url` as returned, or say they are unavailable.
+- Invent or guess a DOI, a `paper_id`, a URL, a page number, or a title.
 - Claim a paper is retracted unless `is_retracted` is true.
-- Treat `found: false` as proof of fabrication. It means the paper is not in the
-  indexed record, which is strong evidence for a mainstream journal article and
-  weak evidence for a book chapter, thesis, technical report, or very recent
-  preprint. Say which case you think it is.
-- Report `citation_count` as a measure of quality. It measures attention, and a
-  recent paper has had less time to accumulate any.
+- Treat `found: false` as proof of fabrication. It is strong evidence for a
+  mainstream journal article and weak evidence for a book chapter, thesis,
+  technical report, conference abstract, or very recent preprint.
+- Report `citation_count` as a measure of quality.
 
 ## Reporting
 
-Give one line per citation, in the user's order, each with a clear verdict:
+One line per citation, in the user's order:
 
 - **Verified** — canonical title, authors, year, venue, DOI.
 - **Retracted** — the same, with the retraction stated first.
 - **Ambiguous** — the candidates, and the question of which was meant.
+- **Mismatch** — what the citation says against the canonical record.
 - **Not found** — what was searched for, and why it may be missing.
 
-Close with the tally: how many verified, how many retracted, how many
-unresolved. If anything was retracted or unresolved, say explicitly that the
-draft needs those references fixed before submission.
+If the user is checking their own draft, say that unresolved or retracted
+references need attention before submission. If the citations come from a
+published paper, an unresolved line means only that it could not be confirmed
+here — not that the published bibliography is wrong.
 
 ## Related workflows
 
+- To check a whole reference list in one call, use `audit-a-bibliography`.
 - To measure a paper's influence rather than its existence, use
   `trace-citation-impact`.
 - To find what else exists on the subject, use `survey-a-research-field`.