prompt-log

What This Skill Does

The prompt-log skill is a specialized utility designed for AI engineers and developers who rely on session logging from tools like Clawdbot, Claude Code, and Codex. At its core, it parses complex, raw .jsonl session files—which are often unreadable to humans—and transforms them into structured, readable Markdown transcripts. By automating the extraction process, this skill ensures that developers can audit interaction histories, document technical progress, or share specific sequences of AI reasoning with team members without manual copy-pasting or complex formatting work.

Installation

To integrate this skill into your environment, ensure you have jq installed on your system, as it is required for JSON processing. On macOS, the script will automatically check for gdate to handle timestamp logic, falling back to standard date if necessary. Install the tool using the following command:

clawhub install openclaw/skills/skills/thesash/prompt-log

Once installed, you can invoke the extraction script directly from your terminal by pointing it at your session files.

Use Cases

• Project Auditing: Review exactly how an AI arrived at a specific architectural decision during a long-running coding session.
• Knowledge Management: Convert transient AI chat sessions into permanent project documentation, capturing both the user queries and the AI's complex code outputs.
• Debugging & Troubleshooting: Isolate specific segments of a session to understand why an AI model failed to produce valid code or hit a rate limit.
• Team Collaboration: Export specific slices of a session (using the --after and --before filters) to demonstrate a complex solution to a colleague without exposing unrelated project context.

Example Prompts

1. "OpenClaw, please run the prompt-log tool on my latest session file in ~/.codex/sessions/ and save the output to my project documentation folder."
2. "Can you extract the interaction history from the clawdbot session I had this morning between 9 AM and 11 AM?"
3. "Run the prompt-log extractor on this session file, but make sure to name the output file feature-x-implementation-notes.md so I can keep my records clean."

Tips & Limitations

• Time Filtering: Always utilize the --after and --before flags when dealing with large, multi-day session files to prevent cluttering your output directory with irrelevant data.
• Environment Dependency: The skill relies heavily on the jq utility. If your extraction fails, confirm that jq is accessible in your system's PATH variable.
• Format Preservation: While the skill outputs clean Markdown, complex multiline code blocks contained within the original .jsonl files are handled best when the source session file is correctly formatted by the AI agent itself. If a session file is corrupted, the extraction output may reflect that structure.