| hylo-core |
contracts, kernel, safety shield, events, trace, registry, config, remote env protocol |
47 |
| hylo-eval |
benchmark contract and splits, scorers, paired statistics and gate tests, evaluation runner, experiments |
8 |
| hylo-agents |
agent composition, reasoners, baseline agents |
15 |
| hylo-memory |
episodic/spatial stores, action memory, lessons, skills |
5 |
| hylo-learn |
learner jobs, instrument calibration and labelling, external data ingestion, learners for runtime components |
9 |
| hylo-evolve |
diagnosis, candidates, validation gate, registry |
11 |
| hylo-catalog |
a read-only query API over the runs directory (runs, episode timelines and frames, paired comparisons, campaigns, artifacts, skills, lessons, memories) |
3 |
| hylo-policies |
one small, weights-free implementation per policy family (retrieval, regression, generative, tokenized, model-based), trained from your own runs |
9 |
| hylo-cli |
the harness subcommands of hylo (run, evolve, learn, ...); hylo itself is in hylo-core |
4 |
| hylo-policy-lerobot |
Hylo adapter that runs LeRobot policies (SmolVLA, pi0, ...) as Agents, usually behind an AgentServer |
8 |
| hylo-policy-openpi |
Hylo adapter that drives an openpi policy server (pi0, pi0-FAST, pi0.5) over its websocket protocol |
2 |
| hylo-reasoner-llm |
client, planner, object choosers, VLM success judge (Ollama native or OpenAI-compatible endpoints) |
6 |
| hylo-proposer-llm |
steering-candidate proposer and lesson reflector (evolve time, never in the control loop) |
3 |
| hylo-data-video |
folders of image sequences and video files (no action labels) |
2 |