synthesize-transcript Read the transcript at: $ARGUMENTS Then analyze how the agent used commands. Your goals are to: 1. Find generalizable learnings — things that would help any agent using this tool — and propose targeted additions to . 2. Surface feature requests and bug fixes for the CLI itself — gaps in functionality, incorrect behavior, or misleading output that should be fixed in code. Process 1. Read the transcript in chunks (it's large). Focus on assistant messages and tool calls/results involving commands. 2. Identify friction and patterns. Look for: - Commands that failed or produc…