AI that gets smarter in YOUR domain — every question compounds.
Online research + local knowledge accumulation + MCP integration for Claude Code & VS Code Copilot.
Demo
Ask → Research → Accumulate → Next question hits local cache instantly
Each round compounds. Knowledge cards have full lifecycle management: draft → reviewed → trusted → stale → deprecated.
Why Scholar Agent?
| ChatGPT / Claude | Obsidian + Plugins | Zotero | Scholar Agent | |
|---|---|---|---|---|
| Domain knowledge accumulates | No — every chat starts fresh | Manual curation | Bibliography only | Automatic — every query compounds |
| Structured research with citations | Sometimes | Manual | Partial | Yes — structured synthesis + confidence scores |
| Academic paper pipeline | Limited | Via plugins | Yes | Search → Score → Analyze → Extract → Recommend |
| Works offline | No | Yes | Partial | Yes — local BM25 index, falls back gracefully |
| Human-readable knowledge base | Chat history | Markdown files | PDF library | Obsidian-compatible Markdown + wiki-links |
| Integrates with your IDE | No | No | Partial | MCP server for Claude Code, VS Code Copilot, OpenCode |
| Knowledge quality gates | None | None | None | Lifecycle management + quality scoring + validation |
Quick Start
Install
pip install py-scholar-agent
Or from source:
git clone https://github.com/zfy465914233/scholar-agent.git
cd scholar-agent
pip install -e .
Setup
scholar-agent init
One command creates data directories, writes config, and registers MCP with Claude Code. You're done.
Modes
| Mode | Command | Data Location | Scope |
|---|---|---|---|
| Global (recommended) | scholar-agent init |
~/scholar/ |
Every project |
| Project-Local | SCHOLAR_HOME=./scholar scholar-agent init |
my-project/scholar/ |
Current project only |
| Docker | docker run -v ~/scholar:/data scholar-agent serve-mcp |
Container volume | Isolated |
What It Does
Knowledge Flywheel
Ask a question → Scholar Agent researches (web + academic APIs) → synthesizes a structured answer with citations → saves as a knowledge card → indexes it with BM25. Next time a similar question comes up, it hits the local index first — fast, accurate, and free.
Academic Research Pipeline
- Paper Search — arXiv, DBLP, Semantic Scholar. Filter by 10+ top conferences
- Smart Scoring — 4-dimensional engine (relevance, recency, popularity, quality)
- Deep Analysis — 20+ section Obsidian-style notes with AI-assisted completion
- Figure Extraction — From arXiv source archives and PDFs
- Daily Recommendations — Automated search, scoring, deduplication
- Paper → Knowledge Card — Feed analyses back into the flywheel
- Keyword Auto-Linking —
[[wiki-links]]across all notes
MCP Tools
Core (always available): query_knowledge · save_research · list_knowledge · capture_answer · ingest_source · build_graph
Academic (set SCHOLAR_ACADEMIC=1): search_papers · search_conf_papers · download_paper · analyze_paper · extract_paper_images · paper_to_card · daily_recommend · link_paper_keywords
CLI Reference
| Command | Description |
|---|---|
scholar-agent init |
One-command setup: data dirs + config + MCP registration |
scholar-agent serve-mcp |
Start the MCP server |
scholar-agent doctor |
Show environment and config diagnostics |
scholar-agent config show |
Show resolved configuration |
scholar-agent install claude --write |
Register MCP with Claude Code |
scholar-agent install vscode --write |
Register MCP with VS Code Copilot |
scholar-agent install opencode --write |
Register MCP with OpenCode |
Configuration
Environment Variables
| Variable | Required | Description |
|---|---|---|
SCHOLAR_ACADEMIC |
No | Set to 1 to enable academic tools |
SCHOLAR_HOME |
No | Override data directory (default: ~/scholar/) |
S2_API_KEY |
No | Semantic Scholar API key (get one free) |
LLM_API_KEY |
No | LLM API key for advanced synthesis pipeline |
Config File
See .scholar.example.json for a full example with comments. Key sections:
knowledge_dir— Knowledge cards directoryindex_path— BM25 search indexacademic.research_interests— Your domains, keywords, arXiv categoriesacademic.scoring— Paper scoring weights
Recommended Workflow
For best paper analysis quality:
- Download:
download_paper("2510.24701", title="Paper Title", domain="LLM") - Extract images:
extract_paper_images("2510.24701") - Deep analysis:
analyze_paper(paper_json)
Downloading the PDF first enables full-text extraction, producing notes with specific data, formulas, and experimental results.
Development
make dev # Install with dev dependencies + pre-commit hooks
make lint # Run ruff + mypy
make test # Run test suite (276 tests, ~7s, fully offline)
make coverage # Run tests with coverage report
make build # Build distribution package
make docker # Build Docker image
See CONTRIBUTING.md for detailed guidelines.
Features at a Glance
- Multi-perspective research — 5 perspectives (academic, technical, applied, contrarian, historical)
- Obsidian compatible — Markdown + YAML frontmatter +
[[wiki-links]] - Knowledge governance — Validate frontmatter, detect orphans, find duplicates, manage lifecycle
- Provider fault tolerance — Each source fails independently; graceful offline fallback
- Cross-platform — macOS, Linux, Windows
License
MIT — see LICENSE.
Metadata
Release files for py-scholar-agent 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| py_scholar_agent-0.1.0.tar.gz | 206.1 kB | With provenance | Details for py_scholar_agent-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| py_scholar_agent-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for py_scholar_agent-0.1.0-py3-none-any.whl |
Total release size: 408.7 kB
Release files / py_scholar_agent-0.1.0.tar.gz
| Download URL | py_scholar_agent-0.1.0.tar.gz |
|---|---|
| Size | 206.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
016bfb09ead7dd3d6d427c1048a66d5c3b6296bf394e49f2f7144919e79afb3c
|
|
BLAKE2b-256 checksum How to use checksums |
d395ed2c217d71d64d57207a74208b392a4e7194694a21564ceb907e0794e3ee
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 1, 2026.
Transparency logRelease files / py_scholar_agent-0.1.0-py3-none-any.whl
| Download URL | py_scholar_agent-0.1.0-py3-none-any.whl |
|---|---|
| Size | 202.6 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
0449e5b30908ff832ce71cd6e82040b398bce3c7ed487f057cbd49dc7c3c60d1
|
|
BLAKE2b-256 checksum How to use checksums |
efa84cecffbd03e4b630bc5e11ff76a5a9853f171f12c13e5146d61276d09047
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 1, 2026.
Transparency logSecurity for py-scholar-agent 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for py_scholar_agent-0.1.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 1, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.