Weekly digest of high-signal AI papers, GitHub momentum, and HuggingFace releases.
Scored by LLM. Delivered via Telegram + email. Built for AI strategists.
arXiv · HuggingFace · GitHub Trending · AlphaSignal
Two-stage filter: keyword pre-screen + Claude Haiku relevance scoring
Author reputation boost · Borderline strategic check · Full synthesis
OpenClaw 🦞 • What It Does • Quick Start • Configuration • Filter Tuning • Architecture • Cron Setup
Market Monitor is an OpenClaw skill. Install it and your agent handles the weekly scan automatically:
https://github.com/yhyatt/MarketMonitor # coming soonOr run manually:
python3 -m market_monitor run --telegram YOUR_CHAT_ID --email [email protected]
|
|
|
|
git clone https://github.com/yhyatt/MarketMonitor.git
cd MarketMonitor
pip install -r requirements.txtexport ANTHROPIC_API_TOKEN=sk-ant-... # Required: LLM scoring
export GOG_KEYRING_PASSWORD=your-password # Required: Gmail delivery
export GITHUB_TOKEN=ghp_... # Optional: higher rate limits (60→5000/hr)Go to alphasignal.ai and subscribe. Create a Gmail label called Digest_sources and apply it to AlphaSignal emails. The collector picks them up automatically.
# Full pipeline: scan + digest
python3 -m market_monitor run --telegram YOUR_CHAT_ID --email [email protected]
# Just scan (no delivery)
python3 -m market_monitor scan
# Just deliver (from stored items)
python3 -m market_monitor digest --telegram YOUR_CHAT_ID --email [email protected]
# Status
python3 -m market_monitor status
# Dry run
python3 -m market_monitor test| Environment Variable | Required | Description |
|---|---|---|
ANTHROPIC_API_TOKEN |
✅ Yes | Anthropic API key for Claude Haiku scoring |
GOG_KEYRING_PASSWORD |
✅ Yes | Password for gog Gmail CLI keyring |
GITHUB_TOKEN |
Optional | GitHub Personal Access Token (raises rate limit from 60→5,000/hr) |
This skill uses gog for Gmail. Make sure gog is authenticated with send scope:
GOG_KEYRING_PASSWORD=your-password gog auth add [email protected]The two-stage filter is tuned for an AI strategist audience. Adjust in market_monitor/filters/:
- Agentic AI, multi-agent systems, orchestration
- Auto-research: autonomous scientific discovery, deep research agents
- Open-source LLMs and frameworks (LLaMA, Mistral, Gemma, DeepSeek, vllm, dspy)
- Paradigm shifts, foundation model positioning
- Human-AI collaboration, enterprise deployment architecture
- Security vulnerabilities — only systemic/novel threats score 8+
- Narrow benchmarks, incremental improvements
- Domain-specific NLP (medical, geographic, etc.)
Researchers on the HIGH_SIGNAL_RESEARCHERS list in scorer.py get +1 to their score, pushing borderline papers (6→7) over the inclusion threshold. Add or remove names to tune.
Items scoring 6–7 after the main scoring pass get a focused second-pass prompt (~150 tokens) asking: "Is there a concrete strategic implication for enterprise AI adoption, market positioning, or paradigm shift?" This can bump a 6→7 (include) or drop a 7→6 (exclude).
┌─────────────────────────────────────────────────────────────────┐
│ COLLECTORS │
│ ┌──────────┐ ┌──────────────┐ ┌────────────┐ ┌───────────┐ │
│ │ arXiv │ │ HuggingFace │ │ GitHub │ │AlphaSignal│ │
│ │ API │ │ API + likes7d│ │ Trending │ │ Gmail │ │
│ └────┬─────┘ └──────┬───────┘ └──────┬─────┘ └─────┬─────┘ │
└───────┼───────────────┼─────────────────┼──────────────┼───────┘
│ │ │ │
▼ ▼ ▼ ▼
┌─────────────────────────────────────────────────────────────────┐
│ FILTERS │
│ ┌────────────────┐ ┌──────────────────┐ ┌─────────────────┐ │
│ │ Keyword Filter │─▶│ LLM Scorer │─▶│ Deduplicator │ │
│ │ (stage 1) │ │ + author boost │ │ │ │
│ │ │ │ + strategic │ │ │ │
│ │ │ │ check (6-7) │ │ │ │
│ └────────────────┘ └──────────────────┘ └─────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ STORE │
│ papers.jsonl · hf_releases.jsonl · github_signals.jsonl │
└─────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ DIGEST │
│ ┌──────────────┐ ┌─────────────────┐ ┌────────────────────┐ │
│ │ Synthesizer │ │ Formatter │ │ Delivery │ │
│ │ (why this │ │ Telegram + HTML │ │ Telegram · Email │ │
│ │ week matters│ │ │ │ │ │
│ └──────────────┘ └─────────────────┘ └────────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
python3 -m market_monitor run --telegram 5553808416 --email [email protected]
Schedule: Sundays 19:00 local time. The isolated agent runs in a clean context.
0 19 * * 0 cd /path/to/workspace && ANTHROPIC_API_TOKEN=sk-ant-... GOG_KEYRING_PASSWORD=... python3 -m market_monitor run --telegram YOUR_CHAT_ID --email [email protected]pip install pytest pytest-mock
python3 -m pytest tests/ -v100 tests, 0 external API calls required (all mocked).
MIT — see LICENSE
Built with 🦞 OpenClaw