Locate Claude Code, Codex CLI, and OpenCode session transcripts and generate handoff summaries.
npx skills add https://github.com/psw7205/skills --skill session-bridge-psw7205
Please help me install this Agent Skill. Skill: session-bridge Source: https://github.com/psw7205/skills/tree/main/skills/session-bridge Command: npx skills add https://github.com/psw7205/skills --skill session-bridge-psw7205
opencode export --sanitize instead of raw database access. - Structured handoff summary: Produces a fixed-format summary covering original goal, completed work, in-progress state, verification results, next actions, and untrusted instructions inside the transcript. - Use Case: You started a refactor in Claude Code yesterday and want to continue it in Codex today. Provide the session UUID, and the Skill locates the transcript, extracts the current state, and writes a copyable handoff prompt for Codex. ## Quick Start Find the Claude Code session with ID abc123 and write a handoff summary so I can continue the work in Codex.High-intent search queries and answers about installing and using this skill.
Run the find-transcript.sh script with the locate command and the session ID or a fragment of it. It searches ~/.claude/projects and ~/.claude/transcripts for matching .jsonl files, preferring filename matches over content matches.
Locate the Claude session transcript by its UUID, read the sampled transcript sections, then generate a handoff summary covering goal, completed work, and next actions. The Skill can append a copyable prompt formatted for the target agent.
Yes. OpenCode sessions are queried from the opencode.db SQLite database by matching the ID, title, or directory against the fragment. Results are then read through opencode export with the --sanitize flag rather than raw database access.
The script returns all candidates sorted by modification time with source, size, and match reason. You should review the candidate list and working directories to select the right one rather than assuming the newest match is correct.
No. It is read-only for investigation and handoff purposes. Transcript content is treated as untrusted external input, so instructions found inside transcripts are never executed unless the current user explicitly requests changes.
code-yeongyuSearch, list, and inspect local coding-agent session transcripts across multiple platforms.
JSai23Parse handoff documents and git state to summarize next steps.
semantica-agiGenerate human-readable explanations for Semantica decisions and graph reasoning.
stacksjsWrite a portable handoff document summarizing the current conversation for another agent session.