Reflect

When a conversation surfaces something worth keeping, this spawns three parallel reviewers (judgment, tooling, and divergent) to mine the transcript for durable patterns, then routes each finding to a concrete skill edit. It handles the full loop: locating your active transcript across three layout formats, running the review subagents with different models and lenses, synthesizing their findings into accepted/rejected/backlog buckets, and either applying trivial edits directly or handing substantive changes to the create-skill workflow. The structural enforcement check is a nice touch: it blocks anything that should live in a linter or runtime guard instead of prose. You approve edits before they land, which matters since skill changes affect every future agent. Use it after complex tasks that worked, dead ends that taught you something, or mid-task corrections you want captured.

skills.sh installs: 4.5K
GitHub stars: 15.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add cursor/plugins --skill reflect --agent claude-code

Installs into .claude/skills of the current project.

Files

Reflect

Mine the current conversation for durable learnings, then route them into skill edits.

Invoke when the user says "reflect" or "/reflect". Skip when the conversation is trivial, off-topic, or already covered by an existing skill the parent followed correctly. One-offs are not learnings.

1. Locate the active transcript

Before fanning out, find this conversation's transcript in the agent-transcripts/ directory the system prompt names. Do not glob across ~/.cursor/projects/*/. That crosses workspace boundaries and reads private chats from unrelated projects.

ls -t <agent-transcripts>/*.jsonl <agent-transcripts>/*/*.jsonl <agent-transcripts>/*/subagents/*.jsonl 2>/dev/null | head -10

That covers the legacy flat (<id>.jsonl), current nested (<id>/<id>.jsonl), and subagent (<parent>/subagents/<child>.jsonl) layouts. For each candidate, read the first JSONL line. Take the file whose message.content[0].text contains the conversation's opening user prompt. If none matches, pass a tight digest of the session instead.