nanameru/claude-transcript-handoff-skill

CC Handoff Skill is a tool that reads the saved conversation history from Claude Code sessions and converts it into a set of human-readable and machine-readable files. Claude Code stores a record of everything that happened in a session, including chat messages, file edits, and tool calls, in JSONL transcript files on your local machine. This skill extracts that data and organizes it into a more useful form. The main use case is handoff: passing the context of a long Claude Code session to another person, to a different AI agent, or to a follow-up session. Without this, someone picking up where a session left off has to either re-read raw transcript files or start fresh without knowing what was already done. The extraction script produces several output files in one pass. These include a summary document covering session statistics and the first and last messages, a file containing the assistant's visible chat messages, a file containing the user's prompts, a spreadsheet-friendly list of every tool call made during the session, machine-readable versions of those tool calls and their outputs, a chronological event timeline, and a file listing where the original transcript files are located. Sensitive content such as API keys, bearer tokens, and JWTs is redacted automatically, though the README notes this is best-effort rather than a guarantee. It also reminds users not to commit the generated handoff files without reviewing them first, since transcripts can contain file paths, command output, and other information that should not be shared publicly. Installation is done through the Vercel Labs Skills CLI, which adds the skill to Codex or Claude Code with a single command. The extraction script can also be run directly with Node.js without installing the skill at all, and accepts arguments to point at a specific working directory or output folder.