agent-lit 0.3.0__tar.gz

agent_lit-0.3.0/.gitignore

@@ -0,0 +1,10 @@
+__pycache__/
+*.py[oc]
+build/
+dist/
+*.egg-info/
+.venv/
+.env
+literature/index/papers.db
+literature/index/papers.db-*
+literature/output/

agent_lit-0.3.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Hangyu Zhou
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

agent_lit-0.3.0/PKG-INFO

@@ -0,0 +1,130 @@
+Metadata-Version: 2.4
+Name: agent-lit
+Version: 0.3.0
+Summary: Persistent memory for AI agents doing literature review. Zero dependencies. SQLite-only.
+Project-URL: Repository, https://github.com/Zhou-Hangyu/alit
+Author: Hangyu Zhou
+License-Expression: MIT
+License-File: LICENSE
+Keywords: agents,alit,literature-review,research,sqlite
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+
+# (-o-) alit
+
+Your AI agent reads papers so you don't have to.
+
+`pip install alit` → zero dependencies, SQLite-only, works with any coding agent.
+
+Got tokens to burn? Let your agent read 50 papers overnight and hand you a synthesis in the morning.
+
+Tell your agent:
+
+```
+Use alit to manage my literature review. See https://github.com/Zhou-Hangyu/alit
+```
+
+## Why alit
+
+Your agent can web-search papers anytime — but forgets everything next session. alit is persistent memory. One agent reads 50 papers, another agent queries that knowledge instantly. Knowledge compounds across sessions.
+
+## How it works
+
+```
+.alit/
+├── papers.db    ← one SQLite file, entire knowledge base
+└── pdfs/        ← auto-downloaded from arXiv
+```
+
+No servers. No API keys. No vector databases. No setup beyond `pip install`.
+
+## Install
+
+```bash
+pip install alit         # or: uv add alit
+alit init                # creates .alit/ in your project
+```
+
+The agent skill auto-installs on first run.
+
+## Set your taste
+
+```bash
+alit taste "I'm into multimodal foundation models and how they learn cross-modal
+representations. Love papers with clean ablations over pure benchmark chasing.
+Especially interested in vision-language grounding and embodied AI."
+```
+
+## Quick start
+
+```bash
+alit add "https://arxiv.org/abs/1706.03762"     # fetches metadata + PDF
+alit import library.bib                          # or dump your Zotero/Mendeley
+alit recommend 5                                 # ranked by your taste
+alit ask "What are the key attention mechanisms?" --depth 2
+```
+
+## Update
+
+```bash
+pip install --upgrade alit
+```
+
+## Commands
+
+| Command | What it does |
+|---------|-------------|
+| `alit init` | Initialize `.alit/` |
+| `alit add <title-or-url>` | Add paper (auto-enriches arXiv, auto-tags) |
+| `alit find <query>` | Search arXiv/S2 for papers by topic |
+| `alit import <file>` | Bulk-add from URL file or BibTeX (.bib) |
+| `alit sync` | Import from remembered BibTeX source (Zotero, etc.) |
+| `alit enrich` | Batch-fetch metadata for papers missing abstracts |
+| `alit search <query>` | BM25 full-text search |
+| `alit recommend [N]` | Reading queue ranked by score |
+| `alit ask <question>` | Cross-paper synthesis via funnel retrieval |
+| `alit read <id>` | Guided reading view |
+| `alit show <id>` | Paper details + citations |
+| `alit list` | List papers |
+| `alit note <id> <text>` | Append reading notes |
+| `alit summarize <id>` | Store summary with model provenance |
+| `alit cite <from> <to>` | Add citation edge |
+| `alit status <id> <s>` | Set reading status |
+| `alit tag <id> <tags>` | Set tags |
+| `alit taste [text]` | Set or show your research taste |
+| `alit progress` | Visual progress dashboard |
+| `alit stats` | Collection overview |
+| `alit orphans` | Find citations to missing papers |
+| `alit attach <id> <pdf>` | Attach local PDF |
+| `alit fetch-pdf <id>` | Download PDF from arXiv |
+| `alit delete <id>` | Remove paper + citations |
+| `alit export [--format X]` | Export as JSON or markdown |
+
+All commands support `--json`.
+
+## Under the hood
+
+- **Search**: BM25 via SQLite FTS5
+- **Ranking**: PageRank on citation graph (pure Python)
+- **Recommendations**: PageRank + recency + taste matching
+- **Synthesis**: multi-stage funnel retrieval (~5K tokens to query 10K papers)
+- **Enrichment**: arXiv API (batched) with Semantic Scholar fallback
+- **Backward compatible**: schema auto-migrates on upgrade
+
+## Development
+
+```bash
+git clone https://github.com/Zhou-Hangyu/alit
+cd alit
+uv sync
+uv run pytest
+```
+
+## License
+
+MIT

agent_lit-0.3.0/README.md

@@ -0,0 +1,113 @@
+# (-o-) alit
+
+Your AI agent reads papers so you don't have to.
+
+`pip install alit` → zero dependencies, SQLite-only, works with any coding agent.
+
+Got tokens to burn? Let your agent read 50 papers overnight and hand you a synthesis in the morning.
+
+Tell your agent:
+
+```
+Use alit to manage my literature review. See https://github.com/Zhou-Hangyu/alit
+```
+
+## Why alit
+
+Your agent can web-search papers anytime — but forgets everything next session. alit is persistent memory. One agent reads 50 papers, another agent queries that knowledge instantly. Knowledge compounds across sessions.
+
+## How it works
+
+```
+.alit/
+├── papers.db    ← one SQLite file, entire knowledge base
+└── pdfs/        ← auto-downloaded from arXiv
+```
+
+No servers. No API keys. No vector databases. No setup beyond `pip install`.
+
+## Install
+
+```bash
+pip install alit         # or: uv add alit
+alit init                # creates .alit/ in your project
+```
+
+The agent skill auto-installs on first run.
+
+## Set your taste
+
+```bash
+alit taste "I'm into multimodal foundation models and how they learn cross-modal
+representations. Love papers with clean ablations over pure benchmark chasing.
+Especially interested in vision-language grounding and embodied AI."
+```
+
+## Quick start
+
+```bash
+alit add "https://arxiv.org/abs/1706.03762"     # fetches metadata + PDF
+alit import library.bib                          # or dump your Zotero/Mendeley
+alit recommend 5                                 # ranked by your taste
+alit ask "What are the key attention mechanisms?" --depth 2
+```
+
+## Update
+
+```bash
+pip install --upgrade alit
+```
+
+## Commands
+
+| Command | What it does |
+|---------|-------------|
+| `alit init` | Initialize `.alit/` |
+| `alit add <title-or-url>` | Add paper (auto-enriches arXiv, auto-tags) |
+| `alit find <query>` | Search arXiv/S2 for papers by topic |
+| `alit import <file>` | Bulk-add from URL file or BibTeX (.bib) |
+| `alit sync` | Import from remembered BibTeX source (Zotero, etc.) |
+| `alit enrich` | Batch-fetch metadata for papers missing abstracts |
+| `alit search <query>` | BM25 full-text search |
+| `alit recommend [N]` | Reading queue ranked by score |
+| `alit ask <question>` | Cross-paper synthesis via funnel retrieval |
+| `alit read <id>` | Guided reading view |
+| `alit show <id>` | Paper details + citations |
+| `alit list` | List papers |
+| `alit note <id> <text>` | Append reading notes |
+| `alit summarize <id>` | Store summary with model provenance |
+| `alit cite <from> <to>` | Add citation edge |
+| `alit status <id> <s>` | Set reading status |
+| `alit tag <id> <tags>` | Set tags |
+| `alit taste [text]` | Set or show your research taste |
+| `alit progress` | Visual progress dashboard |
+| `alit stats` | Collection overview |
+| `alit orphans` | Find citations to missing papers |
+| `alit attach <id> <pdf>` | Attach local PDF |
+| `alit fetch-pdf <id>` | Download PDF from arXiv |
+| `alit delete <id>` | Remove paper + citations |
+| `alit export [--format X]` | Export as JSON or markdown |
+
+All commands support `--json`.
+
+## Under the hood
+
+- **Search**: BM25 via SQLite FTS5
+- **Ranking**: PageRank on citation graph (pure Python)
+- **Recommendations**: PageRank + recency + taste matching
+- **Synthesis**: multi-stage funnel retrieval (~5K tokens to query 10K papers)
+- **Enrichment**: arXiv API (batched) with Semantic Scholar fallback
+- **Backward compatible**: schema auto-migrates on upgrade
+
+## Development
+
+```bash
+git clone https://github.com/Zhou-Hangyu/alit
+cd alit
+uv sync
+uv run pytest
+```
+
+## License
+
+MIT

agent_lit-0.3.0/ROADMAP.md

@@ -0,0 +1,22 @@
+# (-o-) alit roadmap
+
+Features under consideration. Not committed to — needs battle testing first.
+
+## Reading quality
+
+- **Summarization prompts** — Guide agents on how to read papers and write good L4/L2. Different LLMs produce different quality. Needs real-world data before prescribing a format. Would live in `references/reading-guide.md` with just-in-time loading.
+- **Structured L2** — Convention for agents to write L2 as JSON with methods/datasets/results/gaps. Enables experiment agents to query structured data directly.
+
+## Discovery
+
+- **`alit watch`** — Monitor topics, surface new papers matching taste. Self-contained: `alit find` with memory (tracks seen IDs, only shows new).
+- **`alit related <id>`** — Paper-to-paper similarity within collection. BM25 on abstract + shared citations.
+
+## Multi-session
+
+- **`alit diff`** — What changed since last session. New papers, new summaries, status changes. Agents run this at session start to orient.
+- **`alit map`** — Cluster papers by topic, show landscape. Useful at 30+ papers.
+
+## Distribution
+
+- **PyPI publish** — `pip install alit` for everyone, not just source installs.

agent_lit-0.3.0/alit/__main__.py

@@ -0,0 +1,3 @@
+from alit.scripts.lit import main
+
+main()