kindly-software/kdb

Overview

Kindly Debugger (kdb) is an open-source time-travel debugging tool for AI workflows. It records execution traces so developers can step backward through code to inspect state at any prior point.

Best for

Best for
Developers debugging unpredictable AI agent or pipeline behavior

Use cases

• Debugging complex AI pipeline failures by replaying execution history
• Inspecting model state and variable changes across training iterations
• Reproducing intermittent errors in agent or LLM workflows

Notes

Kindly Debugger (kdb) is an open-source time-travel debugging tool for AI workflows. It records execution traces so developers can step backward through code to inspect state at any prior point.

0 stars on GitHub. Last updated 2025-12-12.

Use cases

• Debugging complex AI pipeline failures by replaying execution history
• Inspecting model state and variable changes across training iterations
• Reproducing intermittent errors in agent or LLM workflows

Pros

• Enables deterministic replay of non-deterministic AI runs
• Open-source with no vendor lock-in
• Reduces guesswork by showing exact prior states

Cons

• Zero stars and no community adoption yet
• May have limited documentation or support
• Recording overhead could slow down large-scale workflows

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.