zrwangems/dsh-zrwang-plugin / ars/skills/academic-pipeline
academic-pipeline
Flagged学术研究总编排:研究→写作→诚信检查→评审→返修→复审→终检→定稿的十阶段流水线,每阶段停在用户确认点,含材料护照与 7 类 AI 研究失败模式检查。Triggers: academic pipeline, research to paper, full paper workflow, end-to-end paper | 完整论文流程, 从研究到论文, 端到端写论文, 论文流水线。
_shared/contracts/passport/citation_verification_summary.schema.json
View on GitHub{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"$id": "https://github.com/Imbad0202/academic-research-skills/shared/contracts/passport/citation_verification_summary.schema.json",
"title": "Citation Verification Summary Entry",
"description": "v3.11 #182 Delta 4 derived-view aggregate: one entry per (citation_key, ref_slug) pair, joining the v3.7.3 anchor signal, the v3.9.0/v3.11 triangulation unmatched signals, and the v3.8 audit verdicts into a single 'is this citation verified' view. The read interface for the §3.6 verification_gate contract (#183) and the #184 eval harness. Detection is unconditional (the aggregate is always populated so the signal is always visible, C-V6(e)); only the terminality is policy-gated by terminal_policies.citation_existence (Delta 3 / C-V6, a later batch). lookup_verified uses the v3.11 narrowed-false definition (C-V6(a)): false requires at least one ID-keyed unmatched, never a title-only unmatched. Spec docs/design/2026-05-21-v3.10-182-promote-citation-gate-spec.md §2 Delta 4 + §0(4a) + C-V6(a).",
"type": "object",
"additionalProperties": false,
"required": [
"citation_key",
"ref_slug",
"lookup_verified",
"anchor_present",
"verification_timestamp",
"resolver_outcomes"
],
"properties": {
"citation_key": {
"type": "string",
"description": "Joins to literature_corpus_entry.citation_key."
},
"ref_slug": {
"type": "string",
"description": "Joins to the <!--ref:slug--> markers used in writer prose."
},
"lookup_verified": {
"type": "string",
"enum": [
"true",
"false",
"unresolvable"
],
"description": "3-class existence verdict (a STRING enum, never a JSON boolean). `true` iff >=1 applicable resolver returned status=matched (matched WINS). `false` (v3.11 narrowed, C-V6(a)) iff no applicable resolver is matched AND >=1 applicable resolver is an ID-keyed unmatched (status=unmatched with queried_by=id, i.e. a DOI/arXiv ID that provably fails to resolve = fabrication evidence). `unresolvable` otherwise: every applicable resolver unreachable (total outage), OR every resolver skipped (manual-entry exempt, empty adjudicating set), OR the only negative signals are title-only unmatched (queried_by=title, no resolvable identifier to key on = coverage gap, NOT fabrication). `skipped` outcomes are excluded from classification (applicability/policy, not adjudication). Anti-fabrication bias: a bogus-DOI citation with 3 ID-keyed unmatched + 1 unreachable is `false`, not `unresolvable` (one transient outage does not cancel positive non-existence evidence). The reducer in scripts/citation_verification_summary.py is the single source of truth."
},
"anchor_present": {
"type": "boolean",
"description": "True iff the citation's v3.7.3 anchor marker has kind in {quote, page, section, paragraph} (not none). Read by the #183 epistemic-status gate to know whether the row is anchorless."
},
"verification_timestamp": {
"type": "string",
"format": "date-time",
"description": "ISO-8601 UTC. When >=1 resolver ran, the latest resolver timestamp; when all resolvers were skipped (manual entry) or all unreachable (outage), the attempt timestamp (when the summary was generated). Never null."
},
"resolver_outcomes": {
"type": "object",
"additionalProperties": false,
"description": "Exactly four fields, one per resolver — ALL required (the summary contract is fully-populated; a resolver that did not run carries status=skipped, never an absent key, so a consumer never disambiguates 'absent' from 'skipped'). status is one-of (mutually exclusive). queried_by records what the resolver ACTUALLY queried by (v3.11 #182 Delta 4 / C-V6(a)): the narrowed-false reducer treats an unmatched with queried_by=id as fabrication evidence and queried_by=title as a coverage gap. Downstream consumers needing to distinguish manual exemption from total outage read status directly.",
"required": [
"crossref",
"openalex",
"semantic_scholar",
"arxiv"
],
"properties": {
"crossref": {
"$ref": "#/$defs/resolver_outcome"
},
"openalex": {
"$ref": "#/$defs/resolver_outcome"
},
"semantic_scholar": {
"$ref": "#/$defs/resolver_outcome"
},
"arxiv": {
"$ref": "#/$defs/resolver_outcome"
}
}
},
"cache_age_days": {
"type": "number",
"minimum": 0,
"description": "#541 — citation-level age in days (rounded to 0.1) of the OLDEST live cache row for this citation across resolvers/query forms — deliberately conservative: an unexpired row for an obsolete query form can flag a citation whose latest verification is fresh (accepted advisory-direction false positive). Present only when at least one resolver outcome was served from cache. Never a gate input."
},
"cache_stale_advisory": {
"type": "boolean",
"description": "#541 — true when the emitted (rounded) cache_age_days exceeds ARS_CACHE_STALE_ADVISORY_DAYS (default 30; 0 disables). Advisory-only: surfaces as an ADV-CACHE row at the integrity checkpoint, never gates."
}
},
"$defs": {
"resolver_outcome": {
"type": "object",
"additionalProperties": false,
"required": [
"status",
"queried_by"
],
"properties": {
"status": {
"type": "string",
"enum": [
"matched",
"unmatched",
"unreachable",
"skipped"
],
"description": "matched = resolver returned a match. unmatched = resolver ran and returned no match. unreachable = API outage / timeout. skipped = resolver did not run (manual entry per v3.7.3 exemption, or arXiv resolver on a non-arXiv citation per Delta 1)."
},
"queried_by": {
"type": [
"string",
"null"
],
"enum": [
"id",
"title",
null
],
"description": "v3.11 #182 Delta 4 / C-V6(a): what the resolver actually queried by. REQUIRED (load-bearing for narrowed-false; an absent key would let an ambiguous unmatched silently reduce to unresolvable instead of false). `id` = an exact identifier (DOI or arXiv ID) lookup was attempted (so an `unmatched` here is ID-keyed = fabrication evidence). `title` = only a title search was possible (no resolvable identifier; an `unmatched` here is a coverage gap, reduces to unresolvable not false). `null` for skipped / pure-unreachable outcomes where no meaningful query completed. This reflects resolver EXECUTION, not entry shape (the entry having a DOI does not prove the DOI was queried — it may have been malformed/rejected)."
},
"response_summary": {
"type": [
"string",
"null"
],
"description": "Optional human-readable summary of the resolver response (e.g. matched title, error class). Null when not captured."
}
},
"allOf": [
{
"description": "A ran resolver (matched / unmatched) always knows what it keyed on: queried_by MUST be id or title, never null. This closes the exact ambiguity the narrowed-false reducer can be fooled by (a null-keyed unmatched).",
"if": {
"properties": {
"status": {
"enum": [
"matched",
"unmatched"
]
}
}
},
"then": {
"properties": {
"queried_by": {
"enum": [
"id",
"title"
]
}
}
}
},
{
"description": "A resolver that did not complete a query (skipped / unreachable) MUST carry queried_by=null — there was no meaningful query to attribute.",
"if": {
"properties": {
"status": {
"enum": [
"skipped",
"unreachable"
]
}
}
},
"then": {
"properties": {
"queried_by": {
"const": null
}
}
}
}
]
}
}
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |