How does Aegis decide when to use strict TDD?

It evaluates task risk automatically and chooses strict TDD, lighter verification, or skips TDD when inappropriate.

Can I update Aegis after the first install?

Yes, simply tell the agent to update Aegis or use the explicit aegis:update command; it updates only the current host unless --all is specified.

Is Aegis a background service or daemon?

No, it is a method pack of skills and scripts that the host loads; it does not run independently or grant final authority.

GanyuanRan/Aegis

A baseline-first workflow pack that adds discipline to AI coding agents.

SkillsAgent Skills 528Open source

View on GitHub

Updated 2026-06-16

What is GanyuanRan/Aegis?

Aegis is a runtime-ready method pack that layers structured workflow rules onto existing AI coding agents. It emphasizes reading current project state before risky edits, collecting fresh verification evidence, and documenting whether old code paths are retained or retired.

Agents interact with Aegis through natural language commands that trigger host-specific installation and update scripts. The pack automatically selects light or strict verification modes based on detected risk and supports both automatic and explicit activation.

It targets developers who use AI agents for real software work and want to reduce drift on architecture, ownership, or validation steps without replacing their primary coding host.

What you can build with GanyuanRan/Aegis

High-risk refactors

Forces baseline reading of current code and evidence checks before large structural changes to prevent unintended side effects.

Multi-host environments

Provides consistent workflow rules across different AI coding tools so teams can maintain the same discipline regardless of host.

Verification-heavy tasks

Requires fresh test or review evidence before marking work complete, reducing cases where agents declare success without confirmation.

Install GanyuanRan/Aegis

Quick start

bash tests/e2e/boundary-compliance-check.sh
bash tests/e2e/workflow-quality-check.sh
bash tests/e2e/install-verification-policy-check.sh
bash tests/e2e/layer1-fast-check.sh --host-profile none

1Tell your AI coding agent to read the Aegis GitHub repository and identify your host.
2Instruct the agent to perform the host-specific global installation.
3Restart or reload the host if the agent indicates it is required.
4Run the doctor script from the installed method-pack root to verify configuration.
5Confirm JSON output shows ok, workspaceSupport, and configStatus values before using Aegis.