tjb-tech
5dd8b952ec
A lightweight open-source Python implementation of the Agent Harness architecture. 44x lighter than Claude Code (11K vs 512K lines), 98% core tool coverage. - 43 tools with Pydantic validation and parallel execution - Skills system compatible with anthropics/skills (17+ tested) - Plugin system compatible with claude-code/plugins (12+ tested) - API retry with exponential backoff - Multi-level permissions with path rules - React/Ink TUI with "Oh my Harness!" branding - 114 unit tests + 6 E2E test suites - MIT License
42 lines
1.1 KiB
Python
42 lines
1.1 KiB
Python
"""Tests for the React terminal launcher path."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
from openharness.ui.app import run_repl
|
|
from openharness.ui.react_launcher import build_backend_command
|
|
|
|
|
|
def test_build_backend_command_includes_flags():
|
|
command = build_backend_command(
|
|
cwd="/tmp/demo",
|
|
model="kimi-k2.5",
|
|
base_url="https://api.moonshot.cn/anthropic",
|
|
system_prompt="system",
|
|
api_key="secret",
|
|
)
|
|
assert command[:3] == [command[0], "-m", "openharness"]
|
|
assert "--backend-only" in command
|
|
assert "--cwd" in command
|
|
assert "--model" in command
|
|
assert "--base-url" in command
|
|
assert "--system-prompt" in command
|
|
assert "--api-key" in command
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_run_repl_uses_react_launcher_by_default(monkeypatch):
|
|
seen = {}
|
|
|
|
async def _launch(**kwargs):
|
|
seen.update(kwargs)
|
|
return 0
|
|
|
|
monkeypatch.setattr("openharness.ui.app.launch_react_tui", _launch)
|
|
await run_repl(prompt="hi", cwd="/tmp/demo", model="kimi-k2.5")
|
|
|
|
assert seen["prompt"] == "hi"
|
|
assert seen["cwd"] == "/tmp/demo"
|
|
assert seen["model"] == "kimi-k2.5"
|