skills
Validate bibliographic references in Markdown documents against authoritative sources like journals, preprints, books, webpages, and reports by confirming authors, titles, years, DOIs, arXiv IDs, and ISBNs via web searches. Generate standardized issue reports with title, description, severity, line numbers, and suggested actions for consistent document reviews.
npx claudepluginhub agencyenterprise/draft-detective --plugin skillsComponent Overview
Component Details
Skills (2)
Use this skill whenever you need to report document review issues. It defines the standard issue format — field names, types, severity levels, line-number conventions, and best practices — used across all agent workflows in this project.
Use this skill to validate a bibliographic reference (journal article, preprint, book, webpage, press release, government report, etc.) by searching for it online and comparing the citation to authoritative sources. Invoke when the user asks to check, verify, fact-check, or validate a citation or list of references — confirming author, title, publisher, year, and identifier (DOI / arXiv ID / ISBN / ISSN) against the actual published work.
README
Draft Detective
AI-powered assistant for academic peer review. Built with LangGraph, this tool validates references against claims, flags unsupported assertions, performs literature reviews, and suggests relevant citations — helping reviewers and researchers assess rigor more efficiently.
Note: This project is under active development and not yet ready for production use. The authors will continue to update this repository with the latest work and evaluation results.
Project funded by RAND: https://rand.org/
Goals
The main goal of Draft Detective is to assist and streamline the academic peer review process by reducing manual workload and improving the consistency, transparency, and rigor of evaluations.
Features
- Validate reference against claims: Check the content of each citation to ensure that it backs up the statements made in the text
- Identify unsupported assertions: checks whether any statements need the addition of a reference
- Suggests citations: suggests references to add to statements made in the text
- Literature review: conducts literature review over statements made in the text by searching the web, looking for newer, supporting and contradictory evidences
Architecture
Development
For detailed development setup instructions, see DEVELOPMENT.md.
Deployment
- Railway: See docs/railway-deployment.md for production deployment on Railway
- Kubernetes: See k8s/README.md for Kubernetes/OpenShift deployment
Testing
Tests are organized by type:
tests/unit/- Fast, isolated unit teststests/integration/- Multi-component integration testsevals_inspectai/- LLM-based evaluations using Inspect AI
# Run standard tests (default)
uv run pytest
# Run evaluations (see evals_inspectai/ for available eval suites)
uv run inspect eval evals_inspectai/e2e/reference_validation/reference_validation_e2e.py
License
See LICENSE file
Similar Plugins
academic-writing-agents
Multi-agent orchestrator for academic writing: 12 specialist agents and 30 writing principles for review, research, drafting, polishing, bibliography auditing, and literature surveys.
phd-skills
Research integrity plugin for Claude Code — paper auditing, citation verification, experiment analysis, and methodology-first skills for academic workflows.
verify-references
Verify and validate BibTeX references against CrossRef metadata. Finds uncited entries and flags discrepancies in title/author/journal/volume/pages/year.
academic-research-skills
Production-grade academic research pipeline for Claude Code: research → write → review → revise → finalize. Ships 4 skills (deep-research, academic-paper, academic-paper-reviewer, academic-pipeline) covering 35+ modes, 32-agent ensemble, Material Passport handoff schema, v3.6.7 cross-model audit gate (synthesis + research-architect + report-compiler pattern protection layer), and v3.6.8 generator-evaluator contract for paper drafting.
Imbad0202Help us improve
Share bugs, ideas, or general feedback.