README
¶
ResearchForge
ResearchForge is an open, reproducible research engine for academic literature discovery, systematic review, evidence extraction, and meta-analysis.
The command-line tool is rforge.
Retrieval-first, provenance-first, statistics-first, LLM-assisted.
ResearchForge helps researchers discover papers, map citation graphs, screen literature, extract evidence, run meta-analyses, generate auditable reports, visualize artifacts locally, and maintain a knowledge base of relevant open-source scholarly tooling.
Status
v0.1.0 — pre-alpha. The rforge CLI is functional with 607 passing tests across 29 packages. Core workflows — project workspaces, scholarly source connectors, document parsing, citation graph expansion, screening, evidence extraction, analysis, reporting, and local web GUI — are implemented and test-covered. The project format, CLI surface, and APIs may change before 1.0.
See RESEARCH-FORGE-PRD.md for full product requirements, DEVELOPMENT_PLAN.md for the implementation plan, ROADMAP.md for milestone sequencing, and SKILLS.md for development and agent-usage skills.
Installation
bash <(curl -fsSL https://raw.githubusercontent.com/TrebuchetDynamics/research-forge/main/install.sh)
Or manually:
git clone https://github.com/TrebuchetDynamics/research-forge
cd research-forge
go build -o bin/rforge ./cmd/rforge
Requires Go 1.22+. Optional services (GROBID, OpenSearch, Qdrant, R/metafor) are not required for normal use.
Quickstart
# Create a research project
rforge project create ./my-review --title "High entropy superconductors"
# Search for papers
rforge search --source openalex --query "high entropy superconductors" --from-year 2020
# Import search results into the project library
rforge search import --source openalex --query "high entropy superconductors" \
--pages 3 --project ./my-review
# Check for duplicates
rforge duplicate report --project ./my-review
# Build a report
rforge report build --out ./my-review/report.md
# Launch the local research cockpit
rforge --project ./my-review ui
OSS repository study:
rforge oss add kermitt2/grobid
rforge oss scan --topic "meta-analysis"
rforge oss report --area parsers
Goals
- Cross-source academic search (OpenAlex, Crossref, arXiv, Semantic Scholar, PubMed, Europe PMC, NASA ADS, DOAJ, CORE)
- Legal open-access full-text discovery and acquisition
- PDF and structured-document parsing (GROBID, S2ORC, PaperMage, Anystyle)
- Citation graph expansion and domain mapping
- Deduplication and systematic-review screening workflows
- Evidence extraction linked to exact source passages
- Meta-analysis and statistical reporting (R/metafor, PyMARE)
- Reproducible research packages with full audit trails
- CLI and local Go + HTMX web GUI
Target users
- Academic researchers and graduate students
- Meta-analysis and systematic-review authors
- Independent scientists and research engineers
- R&D teams in physics, engineering, computer science, mathematics, and materials science
Architecture
Research question / domain query
-> Query planner + protocol compiler
-> rforge CLI
-> Scholarly source connectors (OpenAlex, arXiv, Crossref, ...)
-> Local SQLite project store + provenance log
-> Document acquisition + parser adapters (GROBID, S2ORC, ...)
-> Retrieval index (SQLite FTS / OpenSearch / Qdrant)
-> Screening engine + active-learning scaffold
-> Evidence extraction + risk-of-bias
-> Meta-analysis / statistical engine (R/metafor, PyMARE)
-> Report generator + reproducible package exporter
-> Local Go + HTMX research cockpit (rforge ui)
Stack
| Layer | Choice |
|---|---|
| Language | Go |
| CLI | rforge |
| Local web GUI | Go + HTMX (rforge ui) |
| Database | SQLite (PostgreSQL adapter seam planned) |
| Search | SQLite FTS; OpenSearch and Qdrant as optional adapters |
| PDF parsing | GROBID (optional); S2ORC, PaperMage, Anystyle adapter seams |
| Metadata sources | OpenAlex, Crossref, arXiv, Semantic Scholar, PubMed, Europe PMC, NASA ADS, Unpaywall, DOAJ, CORE |
| Meta-analysis | R metafor; PyMARE adapter seam |
Design principle
ResearchForge is a scientific workflow engine, not an opaque AI answer machine. Every output should be able to answer:
- What did we search, and where?
- Where did each paper come from?
- Why was each paper included or excluded?
- What exact source supports each extracted claim?
- What statistical model was run, with which settings?
- Can another researcher reproduce the result from the package?
Agent usage
An agent skill is included for LLM agents running rforge to conduct research and save outputs to a project folder or arbitrary path.
License
MIT License (SPDX: MIT), Copyright (c) 2026 Trebuchet Dynamics. See LICENSE. The license was selected by the repository owner on 2026-06-13; the decision record is tracked in issue #1 and docs/owner-decisions.md.
Decision-gated scope
Local web GUI delivery targets Go + HTMX; the implementation tracker is recorded in issue #2 and ADR 0006. Run make todo-audit to verify that remaining unchecked TODO.md items are covered by owner decisions, make todo-completion-audit for the closeout prompt-to-artifact checklist, or make decisions-markdown for a review-friendly blocker table.
Directories