recovering-sessions

Analyze JSONL agent logs and generate recovery reports with resumption commands.

Updated Dec 17, 2025
One-click install
npx skills add https://github.com/miethe/meatycapture --skill recovering-sessions
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: recovering-sessions
Source: https://github.com/miethe/meatycapture/tree/main/.claude/skills/recovering-sessions
Command: npx skills add https://github.com/miethe/meatycapture --skill recovering-sessions

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a standardized workflow to recover from crashed, failed, or interrupted Claude Code sessions. It analyzes agent logs, verifies on-disk state, and creates a structured recovery report with ready-to-execute resumption commands, helping teams resume work quickly without losing progress.

Core Features & Use Cases

  • Discover, analyze, and correlate recent agent logs to map completed and in-progress work
  • Verify on-disk state against session deliverables to ensure accurate recovery
  • Generate a detailed recovery report with per-task status, verifications, and remediation commands
  • Produce ready-to-run Task() commands to resume interrupted work
  • Integrate with artifact-tracking and Git workflows to record recovery progress and commits

Quick Start

  1. Identify recent logs: run node scripts/find-recent-agents.js for your project
  2. Analyze a log: node scripts/analyze-agent-log.js <path-to-agent-log.jsonl>
  3. Generate a recovery report: node scripts/generate-recovery-report.js
  4. Review recovered work and commit changes
  5. Resume interrupted tasks using the provided Task() commands from the report

Frequently Asked Questions about recovering-sessions

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I recover Claude Code sessions after a crash?

To recover Claude Code sessions after a crash, analyze JSONL agent logs and verify on-disk file state to identify incomplete work. This generates a structured recovery report with ready-to-execute Task() resumption commands.

What is the best way to resume interrupted agent work?

The best way to resume interrupted agent work is to correlate recent agent logs to map in-progress tasks against verified file modifications. This produces ready-to-run resumption commands that integrate with Git workflows to seamlessly continue work.

How does agent log analysis identify completed vs incomplete tasks?

Agent log analysis identifies completed versus incomplete tasks by inspecting JSONL session records to track task progression, then verifying expected file creation and modification states on disk to ensure accurate recovery mapping.

Can I use this session recovery tool for multi-agent crash recovery?

Yes, you can use this session recovery tool for multi-agent crash recovery. It discovers and correlates recent agent logs across multiple sessions to support complex recovery workflows and generate structured handoff reports.

Does Claude Code session recovery integrate with Git workflows?

Claude Code session recovery integrates directly with Git workflows by tracking artifact progress and recording commits. It maps recovered session deliverables against your repository state to ensure accurate remediation before resuming work.

What are the limitations of recovering sessions using agent logs?

Limitations of recovering sessions using agent logs include relying entirely on the availability and integrity of JSONL log files. Recovery accuracy also depends on verifying on-disk file modifications, meaning unlogged state changes may remain undetected.