MindCheck – Analyze your AI coding logs for over-delegation
Scores your hypothesis formation before asking AI, not just token usage.

Streams 10GB+ logs in browser without crashing, beating Splunk on latency.
Backend engineers, DevOps, system administrators analyzing large log files
Splunk · Datadog · ELK Stack
Technical approach: Instead of reading the entire file into RAM (which crashes browsers on 2GB+ files), LV use File.prototype.slice() to read 50KB chunks via FileReader. This keeps memory usage constant at ~10MB regardless of file size. For GZIP files I use the native DecompressionStream API.
What's new in v2.0:
- Command Palette (Ctrl+K) for keyboard-driven navigation
- Multi-Filter: apply multiple include/exclude filters simultaneously with regex support
- Error Aggregation: groups similar errors by pattern, shows occurrence counts
- Split View: two independent panels for comparing different file sections
- Heatmap Navigation: visual error density on minimap (red=error, orange=warn)
- Log Statistics: real-time level distribution with bar charts
- Export Selection: export specific lines as TXT/JSON/CSV
- Dark/Light theme toggle
- Full keyboard shortcut system (Vim-style navigation for search, bookmarks, etc.)
Thanks for checking it out (and very appreciate feedback)!
Scores your hypothesis formation before asking AI, not just token usage.
Yet another coding assessment platform, but this one parses AI agent logs.
Single-binary offline log analytics when Elasticsearch is overkill and ripgrep is too little.
Single-binary speed plus an indexed local DB, terminal heatmaps/graphs and one-command extractors make for a very practical tool when you need answers now. It also bundles unexpected extras — SIGMA rule checks, TF‑IDF and anomaly/AI commands — so it blurs the line between fast grep-replacements and lightweight local analytics. The space is crowded, but the combination of indexing + built-in analytics in a cross-platform Go CLI is a neat, well-realized approach.
Root-cause correlation for Lambda logs—solves real CloudWatch debugging pain better than manual Insights queries.
Deterministic log parsing for K8s crashes before falling back to AI explanations.