V5

[lorenss-m]

Note

Adds v5 with new Environment and Eval APIs (scenarios, integrations, parallel eval), defaults MCP client to FastMCP, replaces OTel with lightweight telemetry, revamps datasets/tools/CLI, and removes RL.

• Core (v5):
  • Introduces hud.environment (connectors, provider integrations, @env.scenario, mock mode) and new hud.eval (EvalContext, run_eval, parallel execution) with v5 Task (keeps LegacyTask).
  • Defaults MCPClient to FastMCP; adds prompt/resource listing, scenario analysis; adds EnvironmentClient.
  • Replaces OTel-based telemetry with lightweight exporter and @instrument; removes hud/otel/* and related code.
• Datasets:
  • New load_tasks/run_dataset pipeline using v5 Task; deprecates old utils and conversion paths; API save/load updated.
• Tools:
  • Adds Coordinate type; updates computer tools (Anthropic/OpenAI/Qwen); makes Jupyter tool optional-deps; grounding refinements.
• CLI/Server:
  • CLI utils tweaks (docker/env detection, metadata w/ scenarios); server uses warnings patch and v5 run_dataset with Task.from_v4.
• Housekeeping:
  • Removes RL module; dependency cleanup ([agents] extra); version bump to 0.5.0; tests updated accordingly.

^{Written by Cursor Bugbot for commit b35738b. This will update automatically on new commits. Configure here.}