Skip to content
Get packs

SkillOpt-Sleep OpenClaw Reference

Reference code for porting a nightly skill-improvement cycle to OpenClaw with DeepSeek; not ready to run

Source Download skill
Repository
microsoft/SkillOpt
Stars
18,813
Licence
MIT
Files
10 files

What it does

This is a contributed reference adaptation, not a finished tool. It shows one way to connect Microsoft's SkillOpt-Sleep cycle (which reviews past task runs and proposes edits to an agent's skill files) to a DeepSeek chat API and sample task files from one contributor's machine. Once installed, the agent gets no new ready-to-use ability; it can help a developer study or port this code. The README and SKILL.md explicitly say not to run or schedule the files unchanged.

Great for

  • Developers studying how SkillOpt-Sleep backends are written
  • Maintainers who want to port the DeepSeek wrapper into a supported OpenClaw integration
  • People comparing custom backends against the shared skillopt_sleep CLI

Skip it if

  • Anyone wanting a plug-and-play nightly optimizer: the SKILL.md says the files contain hardcoded paths and parts are not ported to current interfaces
  • Running the cron script or run_sleep.py on a real machine without rewriting paths and credentials, since they point at one user's home directory and can send task data to a remote API
  • Non-developers: setup requires editing Python, shell and config files and understanding API keys and cron

Try asking

  • “Explain what the SkillOpt-Sleep OpenClaw reference does and which parts are outdated”
  • “List the porting gaps I must fix before the DeepSeek backend works with the current get_backend interface”
  • “Help me replace the hardcoded paths in run_sleep_cron.sh with configurable settings”
  • “Test SkillOpt-Sleep safely using the shared CLI with --backend mock instead of this wrapper”
Step-by-step guide: how to use SkillOpt-Sleep OpenClaw ReferenceInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 10 original files, the MIT licence and an attribution note.

    openclaw.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

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

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip openclaw.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 shared CLI: python -m skillopt_sleep --backend mock to learn the cycle without any API calls or real data
  • Before any real run, replace every absolute path, sample business name and URL, and check that the DeepSeek endpoint and data retention meet your needs
  • Review every staged proposal manually before adoption; auto_adopt is set to false, so this is intended

What's inside

  • config.json
  • README.md
  • run_sleep_cron.sh
  • run_sleep.py
  • SKILL.md
  • skillopt_sleep_openclaw.py
  • slash_sleep.py
  • tests/devops-tasks.json
  • tests/research-cron-tasks.json
  • tests/wiki-tasks.json

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.

Find exactly what
you're looking for