Skip to content
Get packs

SkillOpt-Sleep for Cursor

Learns from recent Cursor sessions and stages reviewed improvements to a project skill

Source Download skill
Repository
microsoft/SkillOpt
Stars
18,813
Licence
MIT
Files
1 file

What it does

It lets Cursor review your recent local coding sessions, find recurring tasks, test candidate improvements to a project skill file against held-out tasks, and stage the best proposal for your review. Nothing live changes until you explicitly adopt it, and it only runs when you ask (or when you explicitly install a schedule). It does not train model weights.

Great for

  • Developers who repeat similar Cursor tasks and want a project skill refined from them
  • Teams wanting a reviewable, validation-gated way to evolve a Cursor project skill
  • Users comfortable running a command-line engine and reviewing diffs before adoption

Skip it if

  • Users who want a fully automatic change to their skills or memory without review, since adoption is a manual step by design
  • Sensitive codebases where session transcripts should not be sent to a model provider, because the cursor, claude, codex, copilot and azure_openai backends make paid provider calls and send replay content
  • Anyone expecting validated results for tool- or repository-dependent tasks, since Cursor tool-aware replay is disabled pending validation

Try asking

  • “Run a status check to see the current SkillOpt-Sleep state and latest staged proposal in my project”
  • “Harvest recent Cursor sessions read-only and show me which recurring tasks were mined”
  • “Do a dry-run with the deterministic mock backend to smoke-test the setup without provider calls”
  • “Run a bounded optimization with the Cursor backend, then review and adopt the staged proposal”
Step-by-step guide: how to use SkillOpt-Sleep for CursorInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    skillopt-sleep.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip skillopt-sleep.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip skillopt-sleep.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Start with the default mock backend and dry-run to verify harvesting before any live model run.
  • Keep runs small with --max-sessions, --max-tasks and --progress, since these limit workload but not total cost.
  • Set evolve_memory to false in ~/.skillopt-sleep/config.json if you do not want CLAUDE.md modified, and review staged proposals before running adopt.

What's inside

  • SKILL.md

Redistributed unmodified from microsoft/SkillOpt under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.