Frequently Asked Questions

Medic uses a deterministic rules engine with 25+ failure patterns. It does not guess — it matches known patterns, shows you the exact evidence from your logs, tells you its confidence level, and recommends a single best fix. ChatGPT gives generic suggestions that may or may not apply to your situation.

Medic works with logs from any AI agent that produces text output. It has optimized detection for OpenClaw, AutoGPT, LangChain, CrewAI, AutoGen, and others. If your agent produces error logs, we can likely diagnose it.

We automatically detect OpenAI, Anthropic, Google, Mistral, Cohere, Groq, Ollama, Azure OpenAI, AWS Bedrock, LiteLLM, and OpenRouter. More providers are added regularly.

No. The entire analysis runs in your web browser. Your logs never leave your device. There are no network requests to any server with your log data. Close the browser tab and everything is gone.

We cover 25+ failure categories including: missing/invalid API keys, authentication failures, gateway timeouts, rate limits, unsupported models, config syntax errors, filesystem permission issues, missing dependencies, container/service failures, provider outages, context length errors, SSL issues, memory errors, network problems, content filter blocks, and more.

We always show a confidence level (high, medium, or low) for each diagnosis. If confidence is low, we will tell you. We never claim certainty without strong evidence from your logs. When we are uncertain, we suggest additional debugging steps.

Not in the current version. Medic is a diagnostic tool — it tells you what broke, why, and what to do. You still apply the fix yourself. Automatic repair capabilities are planned for a future release.

You can paste logs directly from your clipboard or upload files from your local filesystem. We support .txt, .log, .md, .json, .yaml, .yml, .toml, .env, and more. We do not connect to remote servers or monitoring services (yet).

The diagnostic rules engine and pattern matching logic are designed to be transparent and auditable. We plan to open-source the core engine so the community can contribute new rules and failure patterns.

You can reach us via GitHub Issues or email. We actively collect failure patterns from the community to improve our rule library. If you have a log that we should be able to diagnose but cannot, we want to hear about it.