How do I install the Sg skill?

Run npx skillmds@latest add open-horizon-labs/sg in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Sg?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

open-horizon-labs (@open-horizon-labs) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Sg by open-horizon-labs — Agent Skill

Name: Sg
Author: open-horizon-labs

File contents

superego - Metacognitive Oversight

Superego provides metacognitive review for AI sessions, catching premature commitment, scope creep, and misalignment.

All commands invoke the sg CLI. If sg is not installed, show:

sg not installed. Install with:
  brew tap open-horizon-labs/homebrew-tap && brew install superego
  # or: cargo install superego

$sg review

Review staged changes or current work. On-demand metacognitive evaluation.

sg review

$sg init

Initialize superego for this project.

sg init

$sg audit

Review decision history with LLM analysis.

sg audit

$sg history

Query decision history.

sg history

$sg prompt list

List available superego prompts.

sg prompt list

$sg prompt switch

Switch to a different superego prompt.

sg prompt switch <name>

$sg prompt show

Show current prompt content.

sg prompt show

$sg mode

Show current evaluation mode (always or pull).

sg mode

$sg reset

Reset superego state (recovery from corruption).

sg reset

$sg check

Check hooks and auto-update if outdated.

sg check

Protocol

Mode: Pull mode - evaluates only when explicitly requested.

When to use $sg review:

Before committing to a plan or approach
When choosing between alternatives
Before non-trivial implementations
When the task feels complex or uncertain
Before claiming work is "done"

Results:

has_concerns: true = STOP and show feedback to user
has_concerns: false = continue with confidence
skipped: true = proceed normally (this is fine)

Superego is opt-in. Use it for high-stakes decisions, architectural choices, or when you want a second opinion.

---name: sgdescription: Metacognitive oversight. "$sg review" for on-demand evaluation, "$sg init" to set up, "$sg audit" to review decisions.---# superego - Metacognitive OversightSuperego provides metacognitive review for AI sessions, catching premature commitment, scope creep, and misalignment.All commands invoke the `sg` CLI. If sg is not installed, show:```sg not installed. Install with: brew tap open-horizon-labs/homebrew-tap && brew install superego # or: cargo install superego```## $sg reviewReview staged changes or current work. On-demand metacognitive evaluation.```bashsg review```## $sg initInitialize superego for this project.```bashsg init```## $sg auditReview decision history with LLM analysis.```bashsg audit```## $sg historyQuery decision history.```bashsg history```## $sg prompt listList available superego prompts.```bashsg prompt list```## $sg prompt switch <name>Switch to a different superego prompt.```bashsg prompt switch <name>```## $sg prompt showShow current prompt content.```bashsg prompt show```## $sg modeShow current evaluation mode (always or pull).```bashsg mode```## $sg resetReset superego state (recovery from corruption).```bashsg reset```## $sg checkCheck hooks and auto-update if outdated.```bashsg check```---## Protocol**Mode:** Pull mode - evaluates only when explicitly requested.**When to use $sg review:**- Before committing to a plan or approach- When choosing between alternatives- Before non-trivial implementations- When the task feels complex or uncertain- Before claiming work is "done"**Results:**- `has_concerns: true` = STOP and show feedback to user- `has_concerns: false` = continue with confidence- `skipped: true` = proceed normally (this is fine)Superego is opt-in. Use it for high-stakes decisions, architectural choices, or when you want a second opinion.

Frequently asked questions

Sg