Research engine
Skill Tibsfox/gsd-skill-creator/examples/skills/research/research-engine
Introduces a comprehensive agent-based framework for guided software development (GSD)
npx -y skills add Tibsfox/gsd-skill-creator --skill research-engineAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
What its author says it does
Copied from the file, not written here
Autonomous research pipeline — topic to structured documents with HTML/PDF output. Proven at HEL (28 docs, 91K words) and OOPS (9 docs, 20K words) scale.
SKILL.md
2.1 KB, 406 tokens by cl100k_base, as published. Nobody here has run it
Research Engine
Activates when user requests deep research, investigation, or analysis of a topic that will produce multiple structured documents.
Pipeline
TOPIC → DECOMPOSE → PARALLEL RESEARCH → AGGREGATE → STRUCTURE → BUILD → PUBLISH
Stage 1: Decompose
Break the research topic into 4-12 investigation tracks. Each track becomes one document.
Stage 2: Parallel Research
Launch research agents (2-6 in parallel) with specific investigation briefs. Use run_in_background: true for independence.
Stage 3: Aggregate
Collect findings. Check for contradictions. Identify cross-references.
Stage 4: Structure
Organize into a numbered document series with:
- Index page (index.html)
- Individual research documents (research/NN-slug.md)
- Knowledge graph (knowledge-nodes.json)
- Retrospective (retrospective.md)
Stage 5: Build
Convert markdown to HTML and PDF using pandoc + xelatex:
bash build.sh # Uses template.tex and html-template.html
Stage 6: Publish
Commit, tag, merge to main, push, create GitHub release, FTP sync to configured remote host.
Quality Standards
- Minimum 1,500 words per document
- Every claim needs evidence or explicit reasoning
- Cross-references between documents
- Fact-check pass before final publish
- Three output formats: markdown (source), HTML (reading), PDF (sharing)
Project Codes
Each research project gets a 3-4 letter code added to series.js:
- HEL (Helium Supply Chain)
- OOPS (Ecosystem Alignment)
- OPEN (Unsolved Problems)
Build Infrastructure
Templates at the project level:
template.tex— LaTeX template for branded PDFshtml-template.html— Standalone HTML templatebuild.sh— Automated pandoc build script
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.