// the find
kaushikb11/awesome-llm-agents
A curated list of awesome LLM agents frameworks.
An awesome-list of LLM agent frameworks, split into categories from core frameworks to CLI harnesses to safety tooling, with entries stored as individual YAML files that generate the README. It's for developers trying to get a map of the agent-framework landscape before picking one.
Entries live in data/frameworks/*.yml and the README is generated from them, so adding a project is a data change, not a markdown merge-conflict fight. update_metrics.py refreshes stars/language/update-date on a schedule via GitHub Actions, so the numbers aren't stale six months in like most awesome-lists. There's an actual test file for the update script, which is rare for this genre. Inactive/archived projects get moved to their own section with an explicit definition (archived or no push in 12 months) instead of quietly rotting inside the main list.
No visible inclusion bar: 73-star single-maintainer projects with landing-page-style descriptions ('sharpens raw models into principled agents') sit in the same table as LangChain and CrewAI, which undercuts the point of a curated list being a filter. The CLI Agent Harnesses category especially reads like anyone who opened a PR got merged in — a dozen near-identical 'run multiple coding agents at once' wrappers with no differentiation noted. It's a list, not a tool, so there's no code to actually evaluate or run; the only engineering surface is the metrics script. Categorization is a judgment call with no stated criteria, so projects like Mem0 (infra) vs Hindsight (memory) end up in split buckets that aren't obviously distinct.