Aegis
A method pack that makes coding agents architecture-aware: baseline first, evidence checks, drift control and safety across long tasks
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Skills and commands run in the agent environment and read project code
- Installation changes the host's global configuration
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/aegisDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Run one by one in the Claude Code chat
/plugin marketplace add ganyuanran/aegis
/plugin install aegis@aegis-devRead https://github.com/GanyuanRan/Aegis, detect my coding agent host and install Aegis globally with the right guide, then verify from the method-pack root: python scripts/aegis-doctor.py --write-config --json.
Other ways from the author
Read https://github.com/GanyuanRan/Aegis, identify my current AI coding host, and install Aegis globally using the correct host guide. Then verify from the installed method-pack root with python scripts/aegis-doctor.py --write-config --json.
Single-prompt install: the agent finds the host guide and installs the method pack.
This is third-party code. Review the repository files before installing.
What it does
Aegis adds a set of skills and commands that make a coding agent keep the project architecture in mind. Before changes the agent records a baseline, checks its conclusions against facts in the code and tracks drift across long tasks. The method pack installs globally through a host-specific guide and is verified by its own configuration doctor. It works on top of many environments: Claude Code, DeepSeek Harness, GitHub Copilot, Kimi Code CLI and others.
Who it is for. For developers who need the agent to respect architecture on large, long-running tasks.
Good fit when
- The agent loses context and architecture on long tasks
- You need changes checked against facts in the code, not guesses
- You need to track drift between the plan and the real project state
Not a fit when
- Small one-off edits with no architectural risk
- A host that is not among the supported environments
Example request
Record the project baseline and keep edits from breaking the architectureLimitations
Aegis is a method pack, not a platform or background daemon: it makes no final decisions, and user instructions and project rules outrank its guidance. Installation depends on a host-specific guide, and some environments need separate activation. Install completeness must be confirmed with the built-in doctor.
How to disable. Remove the registered method pack through the host install path or delete its directory from the agent's global configuration.
Security check
- Skills and commands run in the agent environment and read project code
- Installation changes the host's global configuration
README in short
The README describes Aegis as a method pack that makes coding agents architecture-aware through a baseline, evidence checks and drift control. Installation is done with a single prompt to the agent or via a host-specific guide, verified through aegis-doctor. Supported hosts include Claude Code, DeepSeek Harness, GitHub Copilot, Kimi Code CLI and several others. It is explicitly not a platform or daemon, but a set of methods on top of the agent. MIT licensed.
FAQ
How do I update Aegis?
After install you can ask the agent to update Aegis in plain language or via aegis:update; updates run through the local path for the current host, with no background auto-updates.
Is this a ready platform?
No, it is a method pack on top of the agent. It does not replace user decisions or project rules.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything