FAQ

General

Q: What is Sisyphus Academica? A self-coordinating pipeline of 20+ specialized agents that helps researchers find papers, generate novel hypotheses, verify citations, and improve writing quality.

Q: Does this require a specific LLM provider? No. Works with any OpenAI-compatible or Anthropic API. Edit config/agent-config.json to switch models.

Q: Do I need OpenCode? The full 10-phase pipeline uses OpenCode’s agent system, but the Python tools (sisyphus search, sisyphus verify) and portable skills work standalone with any agent or CLI.

Q: Is there a demo I can try without installing anything?

git clone https://github.com/argahv/sisyphus-academica.git && cd sisyphus-academica && pip install -e .
sisyphus demo

Installation

Q: How do I install?

git clone https://github.com/argahv/sisyphus-academica.git && cd sisyphus-academica && pip install -e .

Or for the full pipeline: git clone + bash install.sh (requires OpenCode).

Q: Is there a PyPI package? Yes: git clone https://github.com/argahv/sisyphus-academica.git && cd sisyphus-academica && pip install -e .. This installs the CLI tools (sisyphus search, verify, bibtex, demo, configure).

Q: What about the portable skills? The skills/ directory contains 6 novelty engines and 5 reviewer personas as standalone SKILL.md files. Copy them to ~/.claude/skills/ to use with any agent.

Pipeline

Q: How long does a paper take? 30 minutes to 4 hours depending on LLM speed, literature volume, and revision rounds.

Q: What if no novelty engine produces a useful hypothesis? The system reports “no novel angle found” and refuses to write the paper.

Quality

Q: How reliable is citation verification? Every citation checked against 2+ sources (Semantic Scholar AND CrossRef). If neither source finds it, the paper is blocked.

Q: How strict is AI-pattern detection? Very strict. Pattern density < 1 per 2000 words. Zero em dashes allowed. One em dash = automatic failure.

Q: What does “all 10 reviewers must pass” mean? All 10 personas must recommend acceptance. Even one rejection sends the paper back for revision.

Development

Q: How do I run tests?

pip install -e ".[dev]"
python -m pytest tests/ -v

Q: How do I contribute? See CONTRIBUTING.md. Look for good-first-issue labels.