I recovered a bricked LaMetric Time from bare metal with Claude Code
Boots bricked hardware from RAM over USB using FEL protocol—genuine embedded wizardry.
Fix Claude Code 'thinking blocks cannot be modified' API errors caused by streaming interleaving corruption in session JSONL files
The repo does one painful thing well: it inspects SESSION_ID.jsonl, pinpoints where streaming/interleaving corrupted 'thinking' blocks, and offers diagnose, fix (with auto-backup), and nuke fallbacks. It's pragmatic and ruthless — small CLI commands, clear recovery workflow, and rescue-first guidance — but it's narrowly targeted to Claude Code users and won't matter to most people.
Developers and power users of Claude Code / Anthropic Claude who run long-lived sessions and need to recover corrupted session histories
messages.N.content.M: thinking or redacted_thinking blocks in the latest assistant message cannot be modified
When this happens, `--resume` can keep failing and the session feels bricked.
The tool provides:
* diagnose SESSION_ID to find likely corruption points in session JSONL
* fix SESSION_ID for targeted repair (auto-backup before write)
* nuke SESSION_ID as fallback (strip all thinking/redacted_thinking blocks, also auto-backup)
Quick usage:
* claude-session-fix-thinking diagnose SESSION_ID
* claude-session-fix-thinking fix SESSION_ID
* claude --resume SESSION_ID
Repo: https://github.com/miteshashar/claude-code-thinking-blocks-f...
If you every got stuck due to this, you can use this script to resume your session.
Boots bricked hardware from RAM over USB using FEL protocol—genuine embedded wizardry.
Yet another coding assessment platform, but this one parses AI agent logs.
Native macOS UI for Claude Code sessions, but it's a viewer, not a tool—Cursor and IDE plugins already exist.
Deterministic log parsing for K8s crashes before falling back to AI explanations.
Prompt engineering library dressed up as metacognition infrastructure.
Clever spinner-tip hijacking, but it's a curated list glued to someone else's product.