Starstruk Logo MarkStarstruk

v2.0.0.post2

RyanCodrai/turbovecv2.0.0.post2Jul 14, 2026by cuiyuebing

AI Summary

A maintenance patch focusing on governance refinements, sandbox security improvements, and memory system reliability enhancements.

Key Highlights

  • Governance policy engine now bridges frontend tool-guard rules into deep scan paths.
  • Introduction of a restricted token-based Windows sandbox feature.
  • Preservation of venv PATH in shell execution to prevent environment isolation issues.
  • Improved ReMe reliability and CJK embedding safety.
  • Added chat archiving functionality.

New Features

  • Chat archiving
  • Restricted token-based Windows sandbox
  • Governance switch handling for OFF-mode sandbox paths
  • Sandbox venv PATH preservation
  • ReMe reliability improvements

Full Release Notes

## What's Changed
* feat: more sensitive files & allow read global by @weidankong in https://github.com/agentscope-ai/QwenPaw/pull/6067
* chore: bump version to 2.0.0post2 by @cuiyuebing in https://github.com/agentscope-ai/QwenPaw/pull/6070
* test(unit): runtime/security/install regression tests — PR-A4 by @hanson-hex in https://github.com/agentscope-ai/QwenPaw/pull/5813
* feat(chat): update action configuration to disable replace and set ri… by @zhijianma in https://github.com/agentscope-ai/QwenPaw/pull/6072
* chore: fix error in workflow by @rayrayraykk in https://github.com/agentscope-ai/QwenPaw/pull/6073
* fix(doctor): use readiness endpoint in doctor by @wananing in https://github.com/agentscope-ai/QwenPaw/pull/6053
* fix: ASK user for permissionfor sudo by @weidankong in https://github.com/agentscope-ai/QwenPaw/pull/6079
* feat(chat): add chat archiving by @zhijianma in https://github.com/agentscope-ai/QwenPaw/pull/6078
* fix(governance): bridge frontend tool-guard rules into policy deep scan by @vanwaals in https://github.com/agentscope-ai/QwenPaw/pull/6063
* fix(sandbox): preserve venv PATH in shell execution by @fancyboi999 in https://github.com/agentscope-ai/QwenPaw/pull/6081
* fix(goal): clean up stale session after goal completion by @rayrayraykk in https://github.com/agentscope-ai/QwenPaw/pull/6086
* fix: use standard truncation hint for scroll-capped tool results by @niceIrene in https://github.com/agentscope-ai/QwenPaw/pull/5953
* fix(cli): prevent TUI crash when clicking streaming output by @ekzhu in https://github.com/agentscope-ai/QwenPaw/pull/6069
* perf(scroll): batch session history migration writes by @niceIrene in https://github.com/agentscope-ai/QwenPaw/pull/6021
* feat(sandbox): add restricted token based windows sandbox by @ustc-mkh in https://github.com/agentscope-ai/QwenPaw/pull/5931
* feat(approval): use toolSource instead of executionLevel to control Always Allow button by @zhijianma in https://github.com/agentscope-ai/QwenPaw/pull/6095
* fix(goal): reset stop gates on /new and /clear commands by @rayrayraykk in https://github.com/agentscope-ai/QwenPaw/pull/6094
* fix(download_catalog): handle gzip-encoded JSON responses in _fetch_json function by @zhijianma in https://github.com/agentscope-ai/QwenPaw/pull/6106
* feat(memory): improve ReMe reliability, observability, and CJK embedding safety by @jinliyl in https://github.com/agentscope-ai/QwenPaw/pull/6098
* fix(governance): honor sandbox_enabled switch in OFF-mode sandbox path by @vanwaals in https://github.com/agentscope-ai/QwenPaw/pull/6109


**Full Changelog**: https://github.com/agentscope-ai/QwenPaw/compare/v2.0.0.post1...v2.0.0.post2