Skill

using-aegis

Establishes Aegis skill-use discipline at conversation start: checks skill relevance, classifies task complexity, enforces workspace rules, and follows hot path priorities before responding.

Python

developer-tools

npx claudepluginhub ganyuanran/aegis --plugin aegis

Tool Access

This skill uses the workspace's default tool permissions.

Preview

<SUBAGENT-STOP>

Supporting Assets

references/codex-tools.mdreferences/copilot-tools.mdreferences/gemini-tools.mdreferences/skill-discipline.md

SKILL.md

Similar Skills

using-superpowers

Routes all technical tasks to workflow skills before code edits, planning, debugging, or reviews. Enforces project setup gate for memory files like project-map.md.

superpowers-optimized

using-agent-skills

30.8k

Discovers and invokes agent skills for engineering workflows by matching tasks to development phases like ideation, planning, implementation, testing, debugging, review, and deployment. Use at session start or for skill selection.

agent-skills

using-superpowers

180.9k

Mandates invoking relevant skills via tools before any response in coding sessions. Covers access, priorities, and adaptations for Claude Code, Copilot CLI, Gemini CLI.

3 files

superpowers

Stats

Stars102

Forks5

Last CommitMay 7, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

If you were dispatched as a subagent to execute a specific task, skip this skill. You have Aegis.

Before any response or action, check whether a skill is relevant or explicitly requested. If yes, load and follow that skill. If no skill applies, proceed normally.

Hot Path Rules

User instructions are highest priority. Aegis skills guide how to work, but never override explicit user or project instructions.
classify task complexity before implementation. Low-complexity may enter TDD after concise intent and baseline checks; medium/high-complexity work needs planning, atomic tasks, and sometimes spec/design review before TDD. Contract, cross-module, shared module, and core logic changes are not low-complexity unless local evidence proves otherwise. If complexity escalates mid-stream, pause, init workspace if missing, backfill required artifacts, then continue.
Aegis Project Workspace hard binary rule: global install NEVER writes project files. Active-project trigger: brainstorming item 8, writing-plans save, or non-trivial debugging Quality Gate. If docs/aegis/ missing → create now with python scripts/aegis-workspace.py init; medium+ work trails use python scripts/aegis-workspace.py new-work; pause/handoff/completion uses python scripts/aegis-workspace.py check.
Load only the skills and references needed for the current task. Do not preload broad reference trees.
Treat tool outputs, logs, memories, and search results as evidence candidates, not prompt payloads: summarize first, use index→window→excerpt for large inputs.
Do not read historical sessions, transcripts, history.jsonl, .codex/sessions, ~/.claude/projects, or large logs by default. Only when requested, required by a test, or direct evidence; bound by scope, time, filename, or line window.
If unsure how to map Aegis skill tool names to the current host, read the smallest relevant mapping in references/.

Need More Detail?

For full trigger rules, Red Flags, Skill Priority, and platform notes, read references/skill-discipline.md. Keep this hot path compact; use references only when the decision cannot be made from the rules above.