agentscope-ai / agentscope-ai/QwenPaw
Mission parser splits quoted --verify commands
- Dominant language
- Python
- Stars
- 34.9k
- Forks
- 3.1k
- Avg merge
- 1d 15h
- Merged PRs (30d)
- 225
Description
## Problem
`parse_mission_args()` tokenizes Mission arguments with `raw_args.split()`. A quoted verification command containing spaces is split into separate tokens, so the verifier receives an invalid command and part of the command leaks into the task text.
## Reproduction
```text
parse_mission_args('Implement the feature --verify "pytest -q" --max-iterations 7')
# current result:
# {'task_text': 'Implement the feature -q"', 'verify_commands': '"pytest', 'max_iterations': 7}
```
Expected result:
```text
{'task_text': 'Implement the feature', 'verify_commands': 'pytest -q', 'max_iterations': 7}
```
## Proposed direction
Use quote-aware standard-library tokenization for valid Mission arguments, retain the existing whitespace-splitting behavior when quotes are malformed, and show the quoted multi-word form in the Mission help text. Add focused regression coverage for both paths.
Contributor guide
Assessment
This issue has not been assessed yet.