Codex / Claude Code session radar for recovering work context across both platforms.

This skill scans local Codex and Claude Code JSONL histories, builds a unified session index, and helps an Agent answer questions such as:

• What sessions were active recently?
• Which thread should I resume next?
• How do I recover CLI windows after a terminal crash?
• Which Codex or Claude Code session matches this UUID?

This README is for human orientation. Agents should follow SKILL.md.

Do not invoke the bundled script through a relative scripts/... path from an arbitrary project workspace. Resolve the installed skill directory first, then run the script from there.

Windows PowerShell:

$skillDir = "$HOME\.claude\skills\session-review"
$review = Join-Path $skillDir "scripts\session_review.py"

python $review --days 7
python $review --days 1 --action-view
python $review --days 1 --restore-view --shell powershell
python $review --session <uuid>
python $review --json

macOS / Linux / Git Bash:

skill_dir="$HOME/.claude/skills/session-review"
review="$skill_dir/scripts/session_review.py"

python3 "$review" --days 7
python3 "$review" --days 1 --action-view
python3 "$review" --days 1 --restore-view --shell bash
python3 "$review" --session <uuid>
python3 "$review" --json

Use SKILL.md as the Agent-facing instruction file. The main script is:

<skill_dir>/scripts/session_review.py

<skill_dir> means the directory containing this SKILL.md.

Use this skill for panoramic session scanning and recovery queues. Use the session-extract skill for detailed handoff extraction from a specific session. The deprecated claude-code-session-extract and codex-rollout-extract skills are no longer part of the local skill set.