Exploring reliable AI agent systems through small, verifiable open-source contributions.
关注 AI Agent 工程,以问题复现、最小修复与回归测试参与开源。
Python TypeScript Agent tooling Regression testing
- Agent execution — tool calls, configuration fallbacks, and storage integration.
- Data ingestion — consistent limits, URL deduplication, and reader edge cases.
- Reliability — reproduce the behavior, narrow the change, and verify the fix.
A few merged pull requests, with the implementation and review attached.
-
DeerFlow
Backend import resolution, scheduled-task defaults, and MCP path handling.
#7017 · #6185 · #5815 -
Microsoft Agent Framework
Actionable file-write errors for filesystem collisions.
#8710 -
Folio
Respect provider enablement when reading cached data.
#274
Open pull requests
These contributions are still open and have not been merged.

