An OpenEnv-compliant RL environment where an LLM agent must complete a coding task across two sessions with zero shared memory. Session 1 writes a structured handoff note; Session 2 starts cold with only that note.
GET /healthServer health checkGET /docsInteractive API (Swagger)POST /resetStart a new episodePOST /stepExecute an actionWS /wsWebSocket (stateful session)GET /stateCurrent episode stateQuick start:
pip install openenv-core
from client import ContinuityEnvClient, ContinuityAction
with ContinuityEnvClient(
base_url="https://aswini-kumar-cross-session-continuity-env.hf.space"
) as env:
obs = env.reset(difficulty="easy", seed=42)
result = env.step(ContinuityAction(tool="run_tests"))
print(result.observation.output)
Tools:
read_file
write_file
run_tests
write_handoff
parse_handoff
submit