transcript-recovery

VCasecnikovs's avatarfrom VCasecnikovs

Recover full conversation history after context compaction. Use when context window is compacted and Claude needs to access earlier parts of the conversation, when user asks about what was discussed before, or when Claude needs to continue work from a previous session. Reads from /mnt/transcripts/ directory.

⭐0stars🔀0forks📁View on GitHub🕐Updated Dec 26, 2025

When & Why to Use This Skill

The Transcript Recovery skill enables Claude to overcome context window limitations by retrieving full conversation histories from system-stored transcripts. It ensures seamless continuity in long-running tasks, allowing the agent to recall past interactions, search historical logs for specific details, and restore critical context after compaction has occurred.

Use Cases

  • Context Restoration: Automatically recovering full message history after the context window is compacted to ensure the agent doesn't lose track of complex project details.
  • Information Retrieval: Searching through extensive past conversation logs to find specific technical decisions, keywords, or instructions provided in earlier sessions.
  • Workflow Continuity: Resuming work from a previous session by reading historical transcripts to understand the current state of a task.
  • Clean History Export: Extracting and saving clean human-assistant message logs to external files for documentation, auditing, or user review.
nametranscript-recovery
descriptionRecover full conversation history after context compaction. Use when context window is compacted and Claude needs to access earlier parts of the conversation, when user asks about what was discussed before, or when Claude needs to continue work from a previous session. Reads from /mnt/transcripts/ directory.

Transcript Recovery

Recover full conversation history from /mnt/transcripts/ after context compaction.

Quick Start

# Copy script to working directory and run
cp /path/to/skill/scripts/get_transcript.py /home/claude/
python3 /home/claude/get_transcript.py --list

Usage

Command Description
--list List all available transcripts
--all Combine all transcripts chronologically
--messages Extract clean human/assistant messages
--search <query> Search across all transcripts
--file <n> Read specific transcript
--tail <n> Last N lines only
--output <path> Save to file

Common Workflows

After Compaction - Recover Context

python3 /home/claude/get_transcript.py --all --messages

Find Specific Discussion

python3 /home/claude/get_transcript.py --search "topic keyword"

Export Full History for User

python3 /home/claude/get_transcript.py --all --messages -o /mnt/user-data/outputs/history.txt

Notes

  • Transcripts location: /mnt/transcripts/ (read-only)
  • Files contain JSON with unicode escapes - script decodes to UTF-8 automatically
  • journal.txt contains transcript metadata and descriptions
  • Script must be copied to /home/claude/ before execution