Work Session Journaler
Instructions
This skill analyzes a Claude Code session transcript and generates an intelligent summary.
Input Methods
Method 1: Current Session (Manual)
- When invoked during an active session, analyze the current conversation context
Method 2: Transcript File (Hook/Automated)
- Read the transcript file path from the user's message
- Expected format: "Analyze session transcript: /path/to/transcript.jsonl"
- Parse the JSONL transcript file to extract messages and tool usage
Analysis Requirements
-
Read the Transcript: Use the Read tool to load the transcript file (if path provided)
-
Extract Key Information:
- User messages (extract text from both string and structured content)
- Tool calls (Edit, Write, Bash commands, etc.)
- Working directory (cwd)
- Session progression
-
Analyze and Summarize:
- Main Focus: What was the primary goal/task?
- Work Accomplished: What specific changes were made?
- Problems Solved: What issues were addressed?
- Files Modified: Which files were edited/created?
- Git Operations: Commits, PRs, branches
- JIRA/Tickets: Any ticket references
- Tools Used: Key tools and their usage
-
Generate Summary: Create a concise 2-3 paragraph narrative that tells the story of what happened in the session
Output Format
Return a JSON object with the following structure that will be used to create the session summary file:
{
"summary": "Concise narrative summary (2-3 paragraphs)",
"focus": "Main focus/theme of the session",
"files_modified": ["file1.py", "file2.js"],
"git_operations": ["commit", "push", "pr_create"],
"jira_tickets": ["JIRA-123"],
"tools_used": ["Bash", "Edit", "Read"],
"working_directory": "/path/to/project"
}
Important Notes
- Focus on what was accomplished and why, not how
- Identify patterns and themes across the session
- Extract meaningful context from the conversation flow
- Be concise but informative
- If invoked from a hook, the output will be captured and used to create the session summary file