scanned 18d ago·osv · gitleaks · opengrep · picklescan + heuristics·full breakdown in the Security tab ↓
BabyAGI
BabyAGI is the original minimal autonomous AI agent framework. It demonstrated that a small loop — execute task → create new tasks → prioritize tasks → repeat — with vector-based memory could autonomously work toward an objective without per-step human guidance.
Key Features
Task creation loop — after each execution the agent generates follow-up tasks and adds them to the queue
Priority queue — tasks are re-ranked by relevance to the overall objective after each cycle
Vector memory — task results are stored in Chroma/Pinecone and retrieved to give future tasks context
Minimal implementation — the core agent is ~100 lines of Python; easy to understand and extend
Foundation paper — cited in hundreds of agent papers; the starting point for CrewAI, AutoGPT, and many others
Modular — swap the LLM backend, vector store, or execution engine without touching the loop
Quick Start
git clone https://github.com/yoheinakajima/babyagi.git
cd babyagi
pip install -r requirements.txt
OPENAI_API_KEY=sk-... OBJECTIVE="Research the latest advances in fusion energy" python babyagi.py
Install via ai-supply
npx ai-supply add babyagi-autonomous-task-agent
Curated mirror of the open-source BabyAGI (MIT). Get it from the source.
! Security: Review · 8888/100 · grade Bscanned 18d ago
✓ no compromise signals9 risk-surface · 8/20 OWASP controls flagged
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
What this capability can do · med confidence (static)
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
OWASP Top 10 for LLM Applications
⚠LLM03Supply Chainhigh
Vulnerable/compromised dependencies, models or archives in the artifact.