By roomi-fields
Plugin Claude Code anti-hallucination pour la recherche scientifique. Cree des SOTAs (etats de l'art) garantis sans fabrication, audite les SOTAs/articles existants (purge ou warnings), avec un moteur de validation strict (FSM 8 etats + cascade d'acquisition PDF + page 1 anti-homonymie + 19 invariants de coherence). Workflow generique multi-domaines, adapter Obsidian par defaut. Anna's Archive et Sci-Hub strictement opt-in via RESEARCH_ENABLE_SHADOW_LIBS=1.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Troisième passe du pipeline cible refondu. Lance la cascade PDF (10 sources : Crossref OA + arXiv + OpenAlex + Unpaywall + HAL + CORE + archive.org + WebSearch + opt-in Sci-Hub/AA) ciblée sur les refs d'un SOTA donné. Différent de `pipeline run` qui itère sur TOUTES les refs actives — `acquire` ne traite que les refs citées par le SOTA en cours.
Batch audit of every SOTA in the vault. For each SOTA, identifies refs that are page1_validated (technical OK) but not yet sota_cited_confirmed (content audited), and launches audit-sota on those SOTAs. Returns a per-SOTA summary at the end.
Per-citation audit of an academic article (LaTeX or Markdown). Produces RECEIPTS.md classifying each citation as VALID/ADJUST/INVALID/UNVERIFIABLE. Optional --warn inserts inline warnings in a .bak copy of the source.
Audit a SOTA's bibliography. Classify each cited ref by validation state. Optional --purge removes citations to retracted refs from the SOTA (with .bak backup).
Acquire PDFs via the 10-source cascade for a single ref by slug, or a batch filtered by state. Validates page 1 anti-homonymy on each download.
Sub-agent that orchestrates the PDF acquisition cascade for a batch of refs. Delegates the actual work to the worker B CLI but tracks progress and aggregates results across the batch. Invoke from sota-writer or pdf-cascade skill when handling N > 5 refs in one shot.
Sub-agent that parses bibliographic sections and inline citations from SOTA / article text. Takes raw text (a section header + content, or an inline excerpt) and returns structured JSON `[{author, year, title, doi?, venue?, raw}]`. Isolates the LLM extraction from the main agent context. Invoke from the INGEST pipeline whenever a SOTA's bibliographic section or paragraph needs structured citation extraction.
Sub-agent that audits a specific claim against the PDF cited. Invoked by citation-receipts skill for deep PDF↔claim verification. Returns structured verdict (VALID/ADJUST/INVALID/UNVERIFIABLE) with evidence quoted from the source. Isolates the heavy PDF reading from the main agent's context.
Sub-agent that validates whether a downloaded PDF matches the expected metadata (author, title, year). Anti-homonymy check on page 1. Invoke when a manual page 1 verification is needed on an acquired PDF, separate from the cascade's automatic validation.
Sub-agent that performs exhaustive multi-source academic search (paper-search MCP across 22 platforms + optional NotebookLM + optional WebSearch). Returns structured JSON of candidate refs for sota-writer phase A. Invoke when broad literature search is needed without polluting the main agent's context.
Systematic per-citation verification protocol for academic writing. MUST be used whenever writing content that cites sources — papers, blog posts, SOTAs, or any document making factual claims about external work. Triggers (FR + EN) on any writing task involving references, citations, or bibliographic claims, or when the user asks to "verify citations", "check sources", "audit references", "validate bibliography", "produce receipts", "/paper-trail:audit-article", "/paper-trail:receipts". Produces a structured RECEIPTS.md report classifying each citation as VALID / ADJUST / INVALID with reason. Optional --warn mode inserts inline warnings into a .bak copy of the source document. Combined with sota-auditor (upstream existence audit), this skill is the **downstream** per-claim check.
Write academic papers (IMRaD structure) for conferences and journals, with mandatory anti-hallucination citation verification. Triggers (FR + EN) : "écris un paper", "article pour conférence", "rédige pour la conférence", "structure IMRaD", "soumission paper", "write a paper on", "academic article", "/paper-trail:new-paper". Builds on existing SOTAs in the registry — does NOT invent references. Every citation passes through citation-receipts validation before being accepted in the final draft.
Acquire PDFs for bibliographic references via a strict 10-source cascade (Crossref OA → arXiv → OpenAlex → Unpaywall → HAL → CORE → archive.org → WebSearch queue; optionally Sci-Hub + Anna's Archive in opt-in mode). Each acquired PDF is validated against expected author/title/year (page 1 anti-homonymy) before being accepted. Trigger this skill whenever the user wants to download a PDF for a reference, fill a cascade, retry an acquisition, or push a `candidate`/`uid_resolved` ref forward in the FSM. Use also for `/paper-trail:cascade <slug>` and `/paper-trail:reactivate-ocr`. Triggers on French and English phrases: "télécharge le PDF", "lance la cascade", "acquérir les sources", "DL ce papier", "passer en pdf_acquired", "valider page 1", "reprise OCR", "download this paper", "acquire PDFs", "run cascade", "retry acquisition", "advance candidates". The skill never decides whether a citation is truthful — that is the curator's role (sota-auditor skill). It only executes the technical state transitions of the worker B.
Audit the bibliography registry for consistency via 19 mechanical invariants (I1-I19) covering state validity, slug uniqueness, UID format, PDF path normalization, file presence on disk, sha256 integrity, page-1 validation log coherence, state history monotonicity, attempt numbering, blocked-reason presence, citation reciprocity with SOTAs, PDF duplicates, terminal state escapes, RTFM OCR overdue. Trigger this skill whenever the user wants to audit the registry, fix obvious drift, check invariants, or run a coherence sweep. Use for `/paper-trail:doctor`, `/paper-trail:doctor --fix`, `/paper-trail:doctor --correlate-rtfm`. Triggers : "audit registre", "doctor", "vérifier invariants", "fix drift", "check le registre", "audit registry", "run doctor", "auto-fix invariants", "rapport invariants".
Final authority on the truthfulness of references cited in State-of-the-Art reviews and academic papers. Activate this skill whenever the session involves auditing, correcting, or cleaning citations in a SOTA or article, or when responding to verdicts from `pdf-cascade` / `registry-doctor`. Triggers (FR + EN) : "audit SOTA", "corriger SOTA", "vérifier hallucinations", "purger refs", "déclarer une ref véridique ou hallucinée", "audit references", "verify citations", "purge hallucinations", "/paper-trail:audit-sota". Does NOT download (that's pdf-cascade) and does NOT write (that's sota-writer / paper-writer). Decides VRAI vs HALLUCINATION and applies consequences (correct attribution, transition to sota_cited_confirmed, or retract + purge from SOTA). Binary decision, no grey zone.
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Anti-hallucination plugin for academic research in Claude Code. Create literature reviews and papers guaranteed without fabricated citations.
Citation hallucinations in academic writing — fabricated quotes, inverted attributions, misidentified authors — are systemic when the author writes from memory and looks up sources afterward. Retracted papers from major venues have shown the failure mode is structural, not anecdotal.
paper-trail makes the error mechanically impossible by enforcing
a research-first workflow with strict state transitions, automated PDF
acquisition with anti-homonymy validation, and per-citation audit
against the actual source text.
DISCLAIMER.md)RECEIPTS.md classifying each
citation as VALID / ADJUST / INVALID / UNVERIFIABLEpaper-trail's acquisition cascade and search layer cover the major academic indexing and full-text platforms.
| Source | Coverage | Used for |
|---|---|---|
| Crossref | Cross-domain DOI registry, 150M+ records | DOI lookup, metadata, open-access URL |
| arXiv | Preprints (CS, math, physics, q-bio, q-fin, stats, EE) | Full-text PDFs of preprints |
| OpenAlex | 200M+ scholarly works, cross-domain aggregator | Metadata, abstracts, citation graph |
| Unpaywall | 30M+ free full-text articles | OA PDF discovery |
| HAL | French academic repository | Full-text, OA |
| CORE | UK-based aggregator, 200M+ OA records | Full-text fallback |
| archive.org | Digitized books and articles, Internet Archive | Books, older publications, scans |
| Semantic Scholar | AI-curated academic graph, 200M+ papers | Cross-reference, related papers |
| PubMed / PMC / bioRxiv / medRxiv | Biomedical, preprints | Biomedical full-text |
| Zenodo / SSRN / DBLP / DOAJ / BASE / IACR / EuropePMC | Cross-domain | Additional metadata and full-text |
| Source | Coverage | Activation |
|---|---|---|
| Sci-Hub | Paywalled scholarly literature, ~88M papers | RESEARCH_ENABLE_SHADOW_LIBS=1 |
| Anna's Archive | Books and articles, aggregates Library Genesis, Sci-Hub, Z-Library | RESEARCH_ENABLE_SHADOW_LIBS=1 |
Shadow-library activation is explicit and per-session. A
disclaimer prints to stderr on first use. The user is responsible
for legal compliance in their jurisdiction. See
DISCLAIMER.md.
| MCP | Coverage | Status |
|---|---|---|
| paper-search MCP | Unified API over 22 platforms above | Required for SOTA writing — install from git (see INSTALL.md) |
| NotebookLM MCP | Books corpus (Q&A with citations) | Optional — RESEARCH_ENABLE_NOTEBOOKLM=1 |
| RTFM MCP | Local indexed corpus (code, docs, research) | Optional — failure correlation only |
⚠️ Do not install
paper-search-mcpfrom PyPI — the published version (0.1.3) is severely outdated (13 / 63 tools, no OpenAlex, no Crossref, no Semantic Scholar). Use the git HEAD recipe inINSTALL.md.
In a Claude Code session:
/plugin install file:///path/to/paper-trail
Or via marketplace (when published):
/plugin marketplace add roomi-fields
/plugin install paper-trail
RESEARCH_VAULT_PATH is required — the plugin refuses to start
without it. Set it (and any reusable secret like S2_API_KEY) once
in ~/.config/paper-trail/env so every project picks them up
automatically :
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimnpx claudepluginhub roomi-fields/paper-trailAutomate Google NotebookLM at scale. Citation-backed Q&A, full Studio generation (audio, video, infographic, report, presentation, data table), multi-account rotation with auto-reauth, and batch_to_vault for offline RTFM-ingestable answer caching.
Harness-native ECC plugin for engineering teams - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.